{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.43016089672002317,
  "eval_steps": 500,
  "global_step": 52000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.272324936923522e-05,
      "grad_norm": 80462.0859375,
      "learning_rate": 8.080808080808081e-09,
      "loss": 6007.0633,
      "step": 10
    },
    {
      "epoch": 0.00016544649873847045,
      "grad_norm": 360423.46875,
      "learning_rate": 1.6161616161616162e-08,
      "loss": 16793.1375,
      "step": 20
    },
    {
      "epoch": 0.0002481697481077057,
      "grad_norm": 260210.0,
      "learning_rate": 2.4242424242424243e-08,
      "loss": 8807.757,
      "step": 30
    },
    {
      "epoch": 0.0003308929974769409,
      "grad_norm": 333457.03125,
      "learning_rate": 3.2323232323232324e-08,
      "loss": 6780.607,
      "step": 40
    },
    {
      "epoch": 0.0004136162468461761,
      "grad_norm": 240292.671875,
      "learning_rate": 4.040404040404041e-08,
      "loss": 10353.4664,
      "step": 50
    },
    {
      "epoch": 0.0004963394962154114,
      "grad_norm": 205636.09375,
      "learning_rate": 4.8484848484848486e-08,
      "loss": 9239.3687,
      "step": 60
    },
    {
      "epoch": 0.0005790627455846465,
      "grad_norm": 91692.9375,
      "learning_rate": 5.656565656565657e-08,
      "loss": 5793.2141,
      "step": 70
    },
    {
      "epoch": 0.0006617859949538818,
      "grad_norm": 128958.5859375,
      "learning_rate": 6.464646464646465e-08,
      "loss": 7645.5141,
      "step": 80
    },
    {
      "epoch": 0.000744509244323117,
      "grad_norm": 198792.46875,
      "learning_rate": 7.272727272727274e-08,
      "loss": 7540.1,
      "step": 90
    },
    {
      "epoch": 0.0008272324936923522,
      "grad_norm": 143177.3125,
      "learning_rate": 8.080808080808082e-08,
      "loss": 5863.3023,
      "step": 100
    },
    {
      "epoch": 0.0009099557430615875,
      "grad_norm": 69092.4453125,
      "learning_rate": 8.88888888888889e-08,
      "loss": 9028.2383,
      "step": 110
    },
    {
      "epoch": 0.0009926789924308227,
      "grad_norm": 135931.546875,
      "learning_rate": 9.696969696969697e-08,
      "loss": 7231.0602,
      "step": 120
    },
    {
      "epoch": 0.0010754022418000579,
      "grad_norm": 142479.5625,
      "learning_rate": 1.0505050505050506e-07,
      "loss": 4912.673,
      "step": 130
    },
    {
      "epoch": 0.001158125491169293,
      "grad_norm": 310996.5625,
      "learning_rate": 1.1313131313131314e-07,
      "loss": 4637.1719,
      "step": 140
    },
    {
      "epoch": 0.0012408487405385284,
      "grad_norm": 195854.375,
      "learning_rate": 1.2121212121212122e-07,
      "loss": 5191.2895,
      "step": 150
    },
    {
      "epoch": 0.0013235719899077636,
      "grad_norm": 98090.03125,
      "learning_rate": 1.292929292929293e-07,
      "loss": 6609.8664,
      "step": 160
    },
    {
      "epoch": 0.0014062952392769987,
      "grad_norm": 69154.9765625,
      "learning_rate": 1.3737373737373738e-07,
      "loss": 6437.2246,
      "step": 170
    },
    {
      "epoch": 0.001489018488646234,
      "grad_norm": 161934.78125,
      "learning_rate": 1.4545454545454548e-07,
      "loss": 7987.9906,
      "step": 180
    },
    {
      "epoch": 0.0015717417380154692,
      "grad_norm": 72492.5390625,
      "learning_rate": 1.5353535353535356e-07,
      "loss": 5339.8434,
      "step": 190
    },
    {
      "epoch": 0.0016544649873847044,
      "grad_norm": 28318.041015625,
      "learning_rate": 1.6161616161616163e-07,
      "loss": 3778.6113,
      "step": 200
    },
    {
      "epoch": 0.0017371882367539398,
      "grad_norm": 78588.4375,
      "learning_rate": 1.6969696969696974e-07,
      "loss": 5780.0262,
      "step": 210
    },
    {
      "epoch": 0.001819911486123175,
      "grad_norm": 69777.796875,
      "learning_rate": 1.777777777777778e-07,
      "loss": 3939.0852,
      "step": 220
    },
    {
      "epoch": 0.00190263473549241,
      "grad_norm": 26142.01953125,
      "learning_rate": 1.858585858585859e-07,
      "loss": 3253.984,
      "step": 230
    },
    {
      "epoch": 0.0019853579848616455,
      "grad_norm": 77199.8828125,
      "learning_rate": 1.9393939393939395e-07,
      "loss": 3297.4723,
      "step": 240
    },
    {
      "epoch": 0.0020680812342308806,
      "grad_norm": 79269.9921875,
      "learning_rate": 2.0202020202020205e-07,
      "loss": 3044.4252,
      "step": 250
    },
    {
      "epoch": 0.0021508044836001158,
      "grad_norm": 39357.45703125,
      "learning_rate": 2.1010101010101013e-07,
      "loss": 3016.1178,
      "step": 260
    },
    {
      "epoch": 0.002233527732969351,
      "grad_norm": 37892.91015625,
      "learning_rate": 2.181818181818182e-07,
      "loss": 2370.4621,
      "step": 270
    },
    {
      "epoch": 0.002316250982338586,
      "grad_norm": 40799.3984375,
      "learning_rate": 2.2626262626262628e-07,
      "loss": 2173.1178,
      "step": 280
    },
    {
      "epoch": 0.0023989742317078217,
      "grad_norm": 16129.89453125,
      "learning_rate": 2.343434343434344e-07,
      "loss": 2872.4018,
      "step": 290
    },
    {
      "epoch": 0.002481697481077057,
      "grad_norm": 12510.9287109375,
      "learning_rate": 2.4242424242424244e-07,
      "loss": 1385.131,
      "step": 300
    },
    {
      "epoch": 0.002564420730446292,
      "grad_norm": 21687.78515625,
      "learning_rate": 2.505050505050505e-07,
      "loss": 1238.8357,
      "step": 310
    },
    {
      "epoch": 0.002647143979815527,
      "grad_norm": 26216.89453125,
      "learning_rate": 2.585858585858586e-07,
      "loss": 1459.9193,
      "step": 320
    },
    {
      "epoch": 0.0027298672291847623,
      "grad_norm": 54759.859375,
      "learning_rate": 2.666666666666667e-07,
      "loss": 1641.5232,
      "step": 330
    },
    {
      "epoch": 0.0028125904785539974,
      "grad_norm": 11365.4833984375,
      "learning_rate": 2.7474747474747475e-07,
      "loss": 1190.3224,
      "step": 340
    },
    {
      "epoch": 0.002895313727923233,
      "grad_norm": 16354.1533203125,
      "learning_rate": 2.828282828282829e-07,
      "loss": 885.4668,
      "step": 350
    },
    {
      "epoch": 0.002978036977292468,
      "grad_norm": 11645.587890625,
      "learning_rate": 2.9090909090909096e-07,
      "loss": 1400.4568,
      "step": 360
    },
    {
      "epoch": 0.0030607602266617033,
      "grad_norm": 8614.1328125,
      "learning_rate": 2.9898989898989904e-07,
      "loss": 814.0722,
      "step": 370
    },
    {
      "epoch": 0.0031434834760309385,
      "grad_norm": 11007.8818359375,
      "learning_rate": 3.070707070707071e-07,
      "loss": 970.882,
      "step": 380
    },
    {
      "epoch": 0.0032262067254001736,
      "grad_norm": 6238.4267578125,
      "learning_rate": 3.151515151515152e-07,
      "loss": 946.8882,
      "step": 390
    },
    {
      "epoch": 0.003308929974769409,
      "grad_norm": 5815.009765625,
      "learning_rate": 3.2323232323232327e-07,
      "loss": 792.6276,
      "step": 400
    },
    {
      "epoch": 0.003391653224138644,
      "grad_norm": 34222.5390625,
      "learning_rate": 3.3131313131313135e-07,
      "loss": 787.8205,
      "step": 410
    },
    {
      "epoch": 0.0034743764735078795,
      "grad_norm": 4778.3984375,
      "learning_rate": 3.393939393939395e-07,
      "loss": 653.9669,
      "step": 420
    },
    {
      "epoch": 0.0035570997228771147,
      "grad_norm": 2678.25244140625,
      "learning_rate": 3.474747474747475e-07,
      "loss": 793.3054,
      "step": 430
    },
    {
      "epoch": 0.00363982297224635,
      "grad_norm": 1457.694580078125,
      "learning_rate": 3.555555555555556e-07,
      "loss": 630.5825,
      "step": 440
    },
    {
      "epoch": 0.003722546221615585,
      "grad_norm": 5033.21923828125,
      "learning_rate": 3.6363636363636366e-07,
      "loss": 759.3085,
      "step": 450
    },
    {
      "epoch": 0.00380526947098482,
      "grad_norm": 3704.6376953125,
      "learning_rate": 3.717171717171718e-07,
      "loss": 644.2665,
      "step": 460
    },
    {
      "epoch": 0.0038879927203540553,
      "grad_norm": 5654.28662109375,
      "learning_rate": 3.7979797979797987e-07,
      "loss": 710.7244,
      "step": 470
    },
    {
      "epoch": 0.003970715969723291,
      "grad_norm": 1483.4691162109375,
      "learning_rate": 3.878787878787879e-07,
      "loss": 694.1961,
      "step": 480
    },
    {
      "epoch": 0.004053439219092526,
      "grad_norm": 2019.6363525390625,
      "learning_rate": 3.9595959595959597e-07,
      "loss": 573.5805,
      "step": 490
    },
    {
      "epoch": 0.004136162468461761,
      "grad_norm": 1998.4979248046875,
      "learning_rate": 4.040404040404041e-07,
      "loss": 624.7649,
      "step": 500
    },
    {
      "epoch": 0.004218885717830996,
      "grad_norm": 2387.564453125,
      "learning_rate": 4.121212121212122e-07,
      "loss": 626.6972,
      "step": 510
    },
    {
      "epoch": 0.0043016089672002315,
      "grad_norm": 3881.398681640625,
      "learning_rate": 4.2020202020202026e-07,
      "loss": 759.5687,
      "step": 520
    },
    {
      "epoch": 0.004384332216569467,
      "grad_norm": 1924.41845703125,
      "learning_rate": 4.282828282828283e-07,
      "loss": 700.7449,
      "step": 530
    },
    {
      "epoch": 0.004467055465938702,
      "grad_norm": 2266.10693359375,
      "learning_rate": 4.363636363636364e-07,
      "loss": 640.943,
      "step": 540
    },
    {
      "epoch": 0.004549778715307937,
      "grad_norm": 7472.12548828125,
      "learning_rate": 4.444444444444445e-07,
      "loss": 806.1628,
      "step": 550
    },
    {
      "epoch": 0.004632501964677172,
      "grad_norm": 1539.118408203125,
      "learning_rate": 4.5252525252525257e-07,
      "loss": 556.9574,
      "step": 560
    },
    {
      "epoch": 0.004715225214046408,
      "grad_norm": 1867.2452392578125,
      "learning_rate": 4.6060606060606064e-07,
      "loss": 453.8387,
      "step": 570
    },
    {
      "epoch": 0.004797948463415643,
      "grad_norm": 6622.47119140625,
      "learning_rate": 4.686868686868688e-07,
      "loss": 585.511,
      "step": 580
    },
    {
      "epoch": 0.0048806717127848785,
      "grad_norm": 1117.4276123046875,
      "learning_rate": 4.767676767676768e-07,
      "loss": 591.8928,
      "step": 590
    },
    {
      "epoch": 0.004963394962154114,
      "grad_norm": 1792.2279052734375,
      "learning_rate": 4.848484848484849e-07,
      "loss": 659.099,
      "step": 600
    },
    {
      "epoch": 0.005046118211523349,
      "grad_norm": 952.1702880859375,
      "learning_rate": 4.929292929292929e-07,
      "loss": 558.1813,
      "step": 610
    },
    {
      "epoch": 0.005128841460892584,
      "grad_norm": 1103.0098876953125,
      "learning_rate": 5.01010101010101e-07,
      "loss": 488.6359,
      "step": 620
    },
    {
      "epoch": 0.005211564710261819,
      "grad_norm": 2055.734130859375,
      "learning_rate": 5.090909090909092e-07,
      "loss": 611.8516,
      "step": 630
    },
    {
      "epoch": 0.005294287959631054,
      "grad_norm": 1028.85986328125,
      "learning_rate": 5.171717171717172e-07,
      "loss": 568.1,
      "step": 640
    },
    {
      "epoch": 0.005377011209000289,
      "grad_norm": 1175.14892578125,
      "learning_rate": 5.252525252525253e-07,
      "loss": 407.9692,
      "step": 650
    },
    {
      "epoch": 0.005459734458369525,
      "grad_norm": 1459.8382568359375,
      "learning_rate": 5.333333333333335e-07,
      "loss": 496.3047,
      "step": 660
    },
    {
      "epoch": 0.00554245770773876,
      "grad_norm": 3791.671630859375,
      "learning_rate": 5.414141414141415e-07,
      "loss": 492.2279,
      "step": 670
    },
    {
      "epoch": 0.005625180957107995,
      "grad_norm": 1465.2242431640625,
      "learning_rate": 5.494949494949495e-07,
      "loss": 518.7601,
      "step": 680
    },
    {
      "epoch": 0.00570790420647723,
      "grad_norm": 1057.6175537109375,
      "learning_rate": 5.575757575757576e-07,
      "loss": 496.9481,
      "step": 690
    },
    {
      "epoch": 0.005790627455846466,
      "grad_norm": 1338.5599365234375,
      "learning_rate": 5.656565656565658e-07,
      "loss": 493.3083,
      "step": 700
    },
    {
      "epoch": 0.005873350705215701,
      "grad_norm": 1229.9342041015625,
      "learning_rate": 5.737373737373738e-07,
      "loss": 507.9572,
      "step": 710
    },
    {
      "epoch": 0.005956073954584936,
      "grad_norm": 2031.885498046875,
      "learning_rate": 5.818181818181819e-07,
      "loss": 545.9699,
      "step": 720
    },
    {
      "epoch": 0.0060387972039541715,
      "grad_norm": 1317.18408203125,
      "learning_rate": 5.898989898989899e-07,
      "loss": 515.3469,
      "step": 730
    },
    {
      "epoch": 0.006121520453323407,
      "grad_norm": 1839.4345703125,
      "learning_rate": 5.979797979797981e-07,
      "loss": 377.9613,
      "step": 740
    },
    {
      "epoch": 0.006204243702692642,
      "grad_norm": 1210.0321044921875,
      "learning_rate": 6.060606060606061e-07,
      "loss": 465.0116,
      "step": 750
    },
    {
      "epoch": 0.006286966952061877,
      "grad_norm": 2228.900390625,
      "learning_rate": 6.141414141414142e-07,
      "loss": 336.6599,
      "step": 760
    },
    {
      "epoch": 0.006369690201431112,
      "grad_norm": 1690.2906494140625,
      "learning_rate": 6.222222222222223e-07,
      "loss": 402.5558,
      "step": 770
    },
    {
      "epoch": 0.006452413450800347,
      "grad_norm": 1048.852783203125,
      "learning_rate": 6.303030303030304e-07,
      "loss": 460.7069,
      "step": 780
    },
    {
      "epoch": 0.0065351367001695825,
      "grad_norm": 1623.499755859375,
      "learning_rate": 6.383838383838384e-07,
      "loss": 322.7708,
      "step": 790
    },
    {
      "epoch": 0.006617859949538818,
      "grad_norm": 4279.55908203125,
      "learning_rate": 6.464646464646465e-07,
      "loss": 430.4449,
      "step": 800
    },
    {
      "epoch": 0.006700583198908053,
      "grad_norm": 1473.4871826171875,
      "learning_rate": 6.545454545454547e-07,
      "loss": 474.9226,
      "step": 810
    },
    {
      "epoch": 0.006783306448277288,
      "grad_norm": 1544.477783203125,
      "learning_rate": 6.626262626262627e-07,
      "loss": 459.8634,
      "step": 820
    },
    {
      "epoch": 0.006866029697646524,
      "grad_norm": 991.1500854492188,
      "learning_rate": 6.707070707070708e-07,
      "loss": 339.08,
      "step": 830
    },
    {
      "epoch": 0.006948752947015759,
      "grad_norm": 1124.6473388671875,
      "learning_rate": 6.78787878787879e-07,
      "loss": 476.3061,
      "step": 840
    },
    {
      "epoch": 0.007031476196384994,
      "grad_norm": 1139.6890869140625,
      "learning_rate": 6.868686868686869e-07,
      "loss": 504.3425,
      "step": 850
    },
    {
      "epoch": 0.007114199445754229,
      "grad_norm": 1275.02294921875,
      "learning_rate": 6.94949494949495e-07,
      "loss": 431.0188,
      "step": 860
    },
    {
      "epoch": 0.0071969226951234646,
      "grad_norm": 1159.5965576171875,
      "learning_rate": 7.03030303030303e-07,
      "loss": 455.7138,
      "step": 870
    },
    {
      "epoch": 0.0072796459444927,
      "grad_norm": 1401.516357421875,
      "learning_rate": 7.111111111111112e-07,
      "loss": 529.1278,
      "step": 880
    },
    {
      "epoch": 0.007362369193861935,
      "grad_norm": 1474.188720703125,
      "learning_rate": 7.191919191919193e-07,
      "loss": 326.5665,
      "step": 890
    },
    {
      "epoch": 0.00744509244323117,
      "grad_norm": 917.8528442382812,
      "learning_rate": 7.272727272727273e-07,
      "loss": 436.4797,
      "step": 900
    },
    {
      "epoch": 0.007527815692600405,
      "grad_norm": 941.6436157226562,
      "learning_rate": 7.353535353535354e-07,
      "loss": 357.6453,
      "step": 910
    },
    {
      "epoch": 0.00761053894196964,
      "grad_norm": 934.293701171875,
      "learning_rate": 7.434343434343436e-07,
      "loss": 433.9422,
      "step": 920
    },
    {
      "epoch": 0.0076932621913388755,
      "grad_norm": 1530.59814453125,
      "learning_rate": 7.515151515151516e-07,
      "loss": 506.7857,
      "step": 930
    },
    {
      "epoch": 0.007775985440708111,
      "grad_norm": 3313.063720703125,
      "learning_rate": 7.595959595959597e-07,
      "loss": 386.7442,
      "step": 940
    },
    {
      "epoch": 0.007858708690077347,
      "grad_norm": 9505.775390625,
      "learning_rate": 7.676767676767677e-07,
      "loss": 393.8395,
      "step": 950
    },
    {
      "epoch": 0.007941431939446582,
      "grad_norm": 9777.1015625,
      "learning_rate": 7.757575757575758e-07,
      "loss": 444.2824,
      "step": 960
    },
    {
      "epoch": 0.008024155188815817,
      "grad_norm": 952.7528686523438,
      "learning_rate": 7.838383838383839e-07,
      "loss": 463.8062,
      "step": 970
    },
    {
      "epoch": 0.008106878438185052,
      "grad_norm": 1653.4957275390625,
      "learning_rate": 7.919191919191919e-07,
      "loss": 422.011,
      "step": 980
    },
    {
      "epoch": 0.008189601687554287,
      "grad_norm": 1222.734619140625,
      "learning_rate": 8.000000000000001e-07,
      "loss": 370.3142,
      "step": 990
    },
    {
      "epoch": 0.008272324936923522,
      "grad_norm": 1304.0595703125,
      "learning_rate": 8.080808080808082e-07,
      "loss": 423.2224,
      "step": 1000
    },
    {
      "epoch": 0.008355048186292758,
      "grad_norm": 1451.695556640625,
      "learning_rate": 8.161616161616162e-07,
      "loss": 333.3123,
      "step": 1010
    },
    {
      "epoch": 0.008437771435661993,
      "grad_norm": 1295.0885009765625,
      "learning_rate": 8.242424242424244e-07,
      "loss": 395.7991,
      "step": 1020
    },
    {
      "epoch": 0.008520494685031228,
      "grad_norm": 24490.919921875,
      "learning_rate": 8.323232323232324e-07,
      "loss": 484.7744,
      "step": 1030
    },
    {
      "epoch": 0.008603217934400463,
      "grad_norm": 1630.39306640625,
      "learning_rate": 8.404040404040405e-07,
      "loss": 507.0378,
      "step": 1040
    },
    {
      "epoch": 0.008685941183769698,
      "grad_norm": 1444.76708984375,
      "learning_rate": 8.484848484848486e-07,
      "loss": 424.0987,
      "step": 1050
    },
    {
      "epoch": 0.008768664433138933,
      "grad_norm": 9745.6103515625,
      "learning_rate": 8.565656565656566e-07,
      "loss": 356.5501,
      "step": 1060
    },
    {
      "epoch": 0.008851387682508169,
      "grad_norm": 1603.8682861328125,
      "learning_rate": 8.646464646464647e-07,
      "loss": 363.1298,
      "step": 1070
    },
    {
      "epoch": 0.008934110931877404,
      "grad_norm": 1848.8529052734375,
      "learning_rate": 8.727272727272728e-07,
      "loss": 453.0991,
      "step": 1080
    },
    {
      "epoch": 0.009016834181246639,
      "grad_norm": 1721.5032958984375,
      "learning_rate": 8.808080808080808e-07,
      "loss": 352.8759,
      "step": 1090
    },
    {
      "epoch": 0.009099557430615874,
      "grad_norm": 1215.8017578125,
      "learning_rate": 8.88888888888889e-07,
      "loss": 358.2751,
      "step": 1100
    },
    {
      "epoch": 0.00918228067998511,
      "grad_norm": 3627.594482421875,
      "learning_rate": 8.96969696969697e-07,
      "loss": 373.2328,
      "step": 1110
    },
    {
      "epoch": 0.009265003929354344,
      "grad_norm": 1093.44677734375,
      "learning_rate": 9.050505050505051e-07,
      "loss": 305.7944,
      "step": 1120
    },
    {
      "epoch": 0.00934772717872358,
      "grad_norm": 5584.5224609375,
      "learning_rate": 9.131313131313133e-07,
      "loss": 371.0138,
      "step": 1130
    },
    {
      "epoch": 0.009430450428092816,
      "grad_norm": 1449.30419921875,
      "learning_rate": 9.212121212121213e-07,
      "loss": 374.5949,
      "step": 1140
    },
    {
      "epoch": 0.009513173677462052,
      "grad_norm": 1450.5250244140625,
      "learning_rate": 9.292929292929294e-07,
      "loss": 370.9519,
      "step": 1150
    },
    {
      "epoch": 0.009595896926831287,
      "grad_norm": 1153.1729736328125,
      "learning_rate": 9.373737373737376e-07,
      "loss": 311.7418,
      "step": 1160
    },
    {
      "epoch": 0.009678620176200522,
      "grad_norm": 1182.4305419921875,
      "learning_rate": 9.454545454545455e-07,
      "loss": 340.1112,
      "step": 1170
    },
    {
      "epoch": 0.009761343425569757,
      "grad_norm": 2788.3564453125,
      "learning_rate": 9.535353535353536e-07,
      "loss": 402.735,
      "step": 1180
    },
    {
      "epoch": 0.009844066674938992,
      "grad_norm": 1240.61669921875,
      "learning_rate": 9.616161616161617e-07,
      "loss": 430.115,
      "step": 1190
    },
    {
      "epoch": 0.009926789924308227,
      "grad_norm": 7811.59228515625,
      "learning_rate": 9.696969696969698e-07,
      "loss": 294.6839,
      "step": 1200
    },
    {
      "epoch": 0.010009513173677462,
      "grad_norm": 1490.346923828125,
      "learning_rate": 9.77777777777778e-07,
      "loss": 314.4158,
      "step": 1210
    },
    {
      "epoch": 0.010092236423046698,
      "grad_norm": 1153.1387939453125,
      "learning_rate": 9.858585858585858e-07,
      "loss": 422.2901,
      "step": 1220
    },
    {
      "epoch": 0.010174959672415933,
      "grad_norm": 1359.1773681640625,
      "learning_rate": 9.93939393939394e-07,
      "loss": 396.2289,
      "step": 1230
    },
    {
      "epoch": 0.010257682921785168,
      "grad_norm": 1238.34814453125,
      "learning_rate": 1.002020202020202e-06,
      "loss": 432.7112,
      "step": 1240
    },
    {
      "epoch": 0.010340406171154403,
      "grad_norm": 1150.58154296875,
      "learning_rate": 1.01010101010101e-06,
      "loss": 533.3193,
      "step": 1250
    },
    {
      "epoch": 0.010423129420523638,
      "grad_norm": 2107.6708984375,
      "learning_rate": 1.0181818181818183e-06,
      "loss": 386.3309,
      "step": 1260
    },
    {
      "epoch": 0.010505852669892873,
      "grad_norm": 1610.2342529296875,
      "learning_rate": 1.0262626262626264e-06,
      "loss": 378.974,
      "step": 1270
    },
    {
      "epoch": 0.010588575919262109,
      "grad_norm": 10111.8642578125,
      "learning_rate": 1.0343434343434344e-06,
      "loss": 440.8467,
      "step": 1280
    },
    {
      "epoch": 0.010671299168631344,
      "grad_norm": 3038.983642578125,
      "learning_rate": 1.0424242424242426e-06,
      "loss": 531.3744,
      "step": 1290
    },
    {
      "epoch": 0.010754022418000579,
      "grad_norm": 2078.825927734375,
      "learning_rate": 1.0505050505050506e-06,
      "loss": 358.3208,
      "step": 1300
    },
    {
      "epoch": 0.010836745667369814,
      "grad_norm": 1494.41943359375,
      "learning_rate": 1.0585858585858587e-06,
      "loss": 368.8838,
      "step": 1310
    },
    {
      "epoch": 0.01091946891673905,
      "grad_norm": 913.7642211914062,
      "learning_rate": 1.066666666666667e-06,
      "loss": 343.7515,
      "step": 1320
    },
    {
      "epoch": 0.011002192166108284,
      "grad_norm": 1475.224853515625,
      "learning_rate": 1.0747474747474747e-06,
      "loss": 311.1521,
      "step": 1330
    },
    {
      "epoch": 0.01108491541547752,
      "grad_norm": 6028.0986328125,
      "learning_rate": 1.082828282828283e-06,
      "loss": 346.5072,
      "step": 1340
    },
    {
      "epoch": 0.011167638664846755,
      "grad_norm": 1047.4739990234375,
      "learning_rate": 1.090909090909091e-06,
      "loss": 371.3439,
      "step": 1350
    },
    {
      "epoch": 0.01125036191421599,
      "grad_norm": 2157.152099609375,
      "learning_rate": 1.098989898989899e-06,
      "loss": 396.0025,
      "step": 1360
    },
    {
      "epoch": 0.011333085163585225,
      "grad_norm": 1551.3135986328125,
      "learning_rate": 1.1070707070707072e-06,
      "loss": 423.25,
      "step": 1370
    },
    {
      "epoch": 0.01141580841295446,
      "grad_norm": 1080.8287353515625,
      "learning_rate": 1.1151515151515153e-06,
      "loss": 255.2609,
      "step": 1380
    },
    {
      "epoch": 0.011498531662323695,
      "grad_norm": 835.9887084960938,
      "learning_rate": 1.1232323232323233e-06,
      "loss": 284.983,
      "step": 1390
    },
    {
      "epoch": 0.011581254911692932,
      "grad_norm": 774.7077026367188,
      "learning_rate": 1.1313131313131315e-06,
      "loss": 274.2369,
      "step": 1400
    },
    {
      "epoch": 0.011663978161062167,
      "grad_norm": 1325.56982421875,
      "learning_rate": 1.1393939393939395e-06,
      "loss": 797.6222,
      "step": 1410
    },
    {
      "epoch": 0.011746701410431402,
      "grad_norm": 1981.0302734375,
      "learning_rate": 1.1474747474747476e-06,
      "loss": 366.0973,
      "step": 1420
    },
    {
      "epoch": 0.011829424659800638,
      "grad_norm": 2644.739501953125,
      "learning_rate": 1.1555555555555556e-06,
      "loss": 327.9804,
      "step": 1430
    },
    {
      "epoch": 0.011912147909169873,
      "grad_norm": 1665.543212890625,
      "learning_rate": 1.1636363636363638e-06,
      "loss": 425.4298,
      "step": 1440
    },
    {
      "epoch": 0.011994871158539108,
      "grad_norm": 1862.1712646484375,
      "learning_rate": 1.1717171717171719e-06,
      "loss": 324.925,
      "step": 1450
    },
    {
      "epoch": 0.012077594407908343,
      "grad_norm": 1164.2347412109375,
      "learning_rate": 1.1797979797979799e-06,
      "loss": 298.3789,
      "step": 1460
    },
    {
      "epoch": 0.012160317657277578,
      "grad_norm": 1282.484375,
      "learning_rate": 1.187878787878788e-06,
      "loss": 399.5682,
      "step": 1470
    },
    {
      "epoch": 0.012243040906646813,
      "grad_norm": 1316.5758056640625,
      "learning_rate": 1.1959595959595961e-06,
      "loss": 338.0538,
      "step": 1480
    },
    {
      "epoch": 0.012325764156016049,
      "grad_norm": 875.5802001953125,
      "learning_rate": 1.2040404040404042e-06,
      "loss": 308.0409,
      "step": 1490
    },
    {
      "epoch": 0.012408487405385284,
      "grad_norm": 1360.3880615234375,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 292.9588,
      "step": 1500
    },
    {
      "epoch": 0.012491210654754519,
      "grad_norm": 1129.11083984375,
      "learning_rate": 1.2202020202020202e-06,
      "loss": 322.4768,
      "step": 1510
    },
    {
      "epoch": 0.012573933904123754,
      "grad_norm": 2066.356689453125,
      "learning_rate": 1.2282828282828285e-06,
      "loss": 271.5606,
      "step": 1520
    },
    {
      "epoch": 0.012656657153492989,
      "grad_norm": 1669.0103759765625,
      "learning_rate": 1.2363636363636365e-06,
      "loss": 283.8805,
      "step": 1530
    },
    {
      "epoch": 0.012739380402862224,
      "grad_norm": 2613.5205078125,
      "learning_rate": 1.2444444444444445e-06,
      "loss": 376.5511,
      "step": 1540
    },
    {
      "epoch": 0.01282210365223146,
      "grad_norm": 1188.38427734375,
      "learning_rate": 1.2525252525252527e-06,
      "loss": 429.8225,
      "step": 1550
    },
    {
      "epoch": 0.012904826901600695,
      "grad_norm": 1084.84814453125,
      "learning_rate": 1.2606060606060608e-06,
      "loss": 289.73,
      "step": 1560
    },
    {
      "epoch": 0.01298755015096993,
      "grad_norm": 2206.3232421875,
      "learning_rate": 1.268686868686869e-06,
      "loss": 316.7855,
      "step": 1570
    },
    {
      "epoch": 0.013070273400339165,
      "grad_norm": 2453.128662109375,
      "learning_rate": 1.2767676767676768e-06,
      "loss": 304.9484,
      "step": 1580
    },
    {
      "epoch": 0.0131529966497084,
      "grad_norm": 976.6162719726562,
      "learning_rate": 1.2848484848484848e-06,
      "loss": 303.9162,
      "step": 1590
    },
    {
      "epoch": 0.013235719899077635,
      "grad_norm": 1493.1630859375,
      "learning_rate": 1.292929292929293e-06,
      "loss": 356.7861,
      "step": 1600
    },
    {
      "epoch": 0.01331844314844687,
      "grad_norm": 1568.7215576171875,
      "learning_rate": 1.301010101010101e-06,
      "loss": 556.4589,
      "step": 1610
    },
    {
      "epoch": 0.013401166397816106,
      "grad_norm": 2514.0361328125,
      "learning_rate": 1.3090909090909093e-06,
      "loss": 414.0774,
      "step": 1620
    },
    {
      "epoch": 0.01348388964718534,
      "grad_norm": 1089.7327880859375,
      "learning_rate": 1.3171717171717172e-06,
      "loss": 330.1746,
      "step": 1630
    },
    {
      "epoch": 0.013566612896554576,
      "grad_norm": 1226.1904296875,
      "learning_rate": 1.3252525252525254e-06,
      "loss": 408.1962,
      "step": 1640
    },
    {
      "epoch": 0.013649336145923813,
      "grad_norm": 2240.143310546875,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 344.259,
      "step": 1650
    },
    {
      "epoch": 0.013732059395293048,
      "grad_norm": 1064.2305908203125,
      "learning_rate": 1.3414141414141417e-06,
      "loss": 348.617,
      "step": 1660
    },
    {
      "epoch": 0.013814782644662283,
      "grad_norm": 1232.1429443359375,
      "learning_rate": 1.3494949494949497e-06,
      "loss": 406.3739,
      "step": 1670
    },
    {
      "epoch": 0.013897505894031518,
      "grad_norm": 1123.156982421875,
      "learning_rate": 1.357575757575758e-06,
      "loss": 396.9652,
      "step": 1680
    },
    {
      "epoch": 0.013980229143400753,
      "grad_norm": 1280.202880859375,
      "learning_rate": 1.3656565656565657e-06,
      "loss": 391.5166,
      "step": 1690
    },
    {
      "epoch": 0.014062952392769988,
      "grad_norm": 1420.6954345703125,
      "learning_rate": 1.3737373737373738e-06,
      "loss": 437.0295,
      "step": 1700
    },
    {
      "epoch": 0.014145675642139224,
      "grad_norm": 2150.417236328125,
      "learning_rate": 1.381818181818182e-06,
      "loss": 346.5058,
      "step": 1710
    },
    {
      "epoch": 0.014228398891508459,
      "grad_norm": 1029.7105712890625,
      "learning_rate": 1.38989898989899e-06,
      "loss": 326.7657,
      "step": 1720
    },
    {
      "epoch": 0.014311122140877694,
      "grad_norm": 1207.562744140625,
      "learning_rate": 1.3979797979797982e-06,
      "loss": 279.2056,
      "step": 1730
    },
    {
      "epoch": 0.014393845390246929,
      "grad_norm": 919.50244140625,
      "learning_rate": 1.406060606060606e-06,
      "loss": 264.9025,
      "step": 1740
    },
    {
      "epoch": 0.014476568639616164,
      "grad_norm": 1641.6435546875,
      "learning_rate": 1.4141414141414143e-06,
      "loss": 239.4691,
      "step": 1750
    },
    {
      "epoch": 0.0145592918889854,
      "grad_norm": 1205.822998046875,
      "learning_rate": 1.4222222222222223e-06,
      "loss": 333.4085,
      "step": 1760
    },
    {
      "epoch": 0.014642015138354635,
      "grad_norm": 1508.9305419921875,
      "learning_rate": 1.4303030303030306e-06,
      "loss": 300.8089,
      "step": 1770
    },
    {
      "epoch": 0.01472473838772387,
      "grad_norm": 1087.9306640625,
      "learning_rate": 1.4383838383838386e-06,
      "loss": 292.6564,
      "step": 1780
    },
    {
      "epoch": 0.014807461637093105,
      "grad_norm": 2015.249755859375,
      "learning_rate": 1.4464646464646464e-06,
      "loss": 277.2897,
      "step": 1790
    },
    {
      "epoch": 0.01489018488646234,
      "grad_norm": 4033.002685546875,
      "learning_rate": 1.4545454545454546e-06,
      "loss": 336.2396,
      "step": 1800
    },
    {
      "epoch": 0.014972908135831575,
      "grad_norm": 1528.888916015625,
      "learning_rate": 1.4626262626262627e-06,
      "loss": 310.2283,
      "step": 1810
    },
    {
      "epoch": 0.01505563138520081,
      "grad_norm": 1040.3310546875,
      "learning_rate": 1.470707070707071e-06,
      "loss": 296.3051,
      "step": 1820
    },
    {
      "epoch": 0.015138354634570046,
      "grad_norm": 3019.808837890625,
      "learning_rate": 1.478787878787879e-06,
      "loss": 278.5342,
      "step": 1830
    },
    {
      "epoch": 0.01522107788393928,
      "grad_norm": 1161.161376953125,
      "learning_rate": 1.4868686868686872e-06,
      "loss": 357.7166,
      "step": 1840
    },
    {
      "epoch": 0.015303801133308516,
      "grad_norm": 2254.221435546875,
      "learning_rate": 1.494949494949495e-06,
      "loss": 325.2632,
      "step": 1850
    },
    {
      "epoch": 0.015386524382677751,
      "grad_norm": 3298.8662109375,
      "learning_rate": 1.5030303030303032e-06,
      "loss": 340.6655,
      "step": 1860
    },
    {
      "epoch": 0.015469247632046986,
      "grad_norm": 1297.3563232421875,
      "learning_rate": 1.5111111111111112e-06,
      "loss": 327.4712,
      "step": 1870
    },
    {
      "epoch": 0.015551970881416221,
      "grad_norm": 1163.013671875,
      "learning_rate": 1.5191919191919195e-06,
      "loss": 302.4008,
      "step": 1880
    },
    {
      "epoch": 0.015634694130785456,
      "grad_norm": 3574.491943359375,
      "learning_rate": 1.5272727272727275e-06,
      "loss": 304.6065,
      "step": 1890
    },
    {
      "epoch": 0.015717417380154693,
      "grad_norm": 953.6083984375,
      "learning_rate": 1.5353535353535353e-06,
      "loss": 283.16,
      "step": 1900
    },
    {
      "epoch": 0.015800140629523927,
      "grad_norm": 579.375732421875,
      "learning_rate": 1.5434343434343435e-06,
      "loss": 294.0983,
      "step": 1910
    },
    {
      "epoch": 0.015882863878893164,
      "grad_norm": 904.5838012695312,
      "learning_rate": 1.5515151515151516e-06,
      "loss": 277.4686,
      "step": 1920
    },
    {
      "epoch": 0.015965587128262397,
      "grad_norm": 878.8187255859375,
      "learning_rate": 1.5595959595959598e-06,
      "loss": 348.8653,
      "step": 1930
    },
    {
      "epoch": 0.016048310377631634,
      "grad_norm": 1161.1363525390625,
      "learning_rate": 1.5676767676767678e-06,
      "loss": 374.7675,
      "step": 1940
    },
    {
      "epoch": 0.016131033627000867,
      "grad_norm": 1667.3831787109375,
      "learning_rate": 1.5757575757575759e-06,
      "loss": 359.4971,
      "step": 1950
    },
    {
      "epoch": 0.016213756876370104,
      "grad_norm": 2197.3330078125,
      "learning_rate": 1.5838383838383839e-06,
      "loss": 367.9104,
      "step": 1960
    },
    {
      "epoch": 0.016296480125739338,
      "grad_norm": 1425.8123779296875,
      "learning_rate": 1.5919191919191921e-06,
      "loss": 367.0379,
      "step": 1970
    },
    {
      "epoch": 0.016379203375108575,
      "grad_norm": 4604.79052734375,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 462.412,
      "step": 1980
    },
    {
      "epoch": 0.016461926624477808,
      "grad_norm": 1806.8238525390625,
      "learning_rate": 1.6080808080808084e-06,
      "loss": 350.7875,
      "step": 1990
    },
    {
      "epoch": 0.016544649873847045,
      "grad_norm": 1519.6781005859375,
      "learning_rate": 1.6161616161616164e-06,
      "loss": 272.1075,
      "step": 2000
    },
    {
      "epoch": 0.01662737312321628,
      "grad_norm": 2001.111572265625,
      "learning_rate": 1.6242424242424242e-06,
      "loss": 264.7772,
      "step": 2010
    },
    {
      "epoch": 0.016710096372585515,
      "grad_norm": 2383.3994140625,
      "learning_rate": 1.6323232323232325e-06,
      "loss": 293.2167,
      "step": 2020
    },
    {
      "epoch": 0.016792819621954752,
      "grad_norm": 670.5611572265625,
      "learning_rate": 1.6404040404040405e-06,
      "loss": 379.9431,
      "step": 2030
    },
    {
      "epoch": 0.016875542871323986,
      "grad_norm": 1432.4300537109375,
      "learning_rate": 1.6484848484848487e-06,
      "loss": 362.749,
      "step": 2040
    },
    {
      "epoch": 0.016958266120693222,
      "grad_norm": 6461.77880859375,
      "learning_rate": 1.6565656565656567e-06,
      "loss": 407.5534,
      "step": 2050
    },
    {
      "epoch": 0.017040989370062456,
      "grad_norm": 1288.83544921875,
      "learning_rate": 1.6646464646464648e-06,
      "loss": 315.7526,
      "step": 2060
    },
    {
      "epoch": 0.017123712619431693,
      "grad_norm": 907.4447021484375,
      "learning_rate": 1.6727272727272728e-06,
      "loss": 292.4799,
      "step": 2070
    },
    {
      "epoch": 0.017206435868800926,
      "grad_norm": 1921.0576171875,
      "learning_rate": 1.680808080808081e-06,
      "loss": 297.3491,
      "step": 2080
    },
    {
      "epoch": 0.017289159118170163,
      "grad_norm": 1620.2481689453125,
      "learning_rate": 1.688888888888889e-06,
      "loss": 291.1318,
      "step": 2090
    },
    {
      "epoch": 0.017371882367539396,
      "grad_norm": 892.2626342773438,
      "learning_rate": 1.6969696969696973e-06,
      "loss": 337.7454,
      "step": 2100
    },
    {
      "epoch": 0.017454605616908633,
      "grad_norm": 1724.4583740234375,
      "learning_rate": 1.705050505050505e-06,
      "loss": 386.2352,
      "step": 2110
    },
    {
      "epoch": 0.017537328866277867,
      "grad_norm": 1373.0406494140625,
      "learning_rate": 1.7131313131313131e-06,
      "loss": 312.9512,
      "step": 2120
    },
    {
      "epoch": 0.017620052115647104,
      "grad_norm": 1802.06884765625,
      "learning_rate": 1.7212121212121214e-06,
      "loss": 392.6038,
      "step": 2130
    },
    {
      "epoch": 0.017702775365016337,
      "grad_norm": 1823.2747802734375,
      "learning_rate": 1.7292929292929294e-06,
      "loss": 276.9635,
      "step": 2140
    },
    {
      "epoch": 0.017785498614385574,
      "grad_norm": 2303.46484375,
      "learning_rate": 1.7373737373737376e-06,
      "loss": 363.1688,
      "step": 2150
    },
    {
      "epoch": 0.017868221863754807,
      "grad_norm": 939.9644165039062,
      "learning_rate": 1.7454545454545456e-06,
      "loss": 371.3494,
      "step": 2160
    },
    {
      "epoch": 0.017950945113124044,
      "grad_norm": 803.9099731445312,
      "learning_rate": 1.7535353535353537e-06,
      "loss": 325.9621,
      "step": 2170
    },
    {
      "epoch": 0.018033668362493278,
      "grad_norm": 1147.0989990234375,
      "learning_rate": 1.7616161616161617e-06,
      "loss": 339.9603,
      "step": 2180
    },
    {
      "epoch": 0.018116391611862515,
      "grad_norm": 1110.0518798828125,
      "learning_rate": 1.76969696969697e-06,
      "loss": 311.3122,
      "step": 2190
    },
    {
      "epoch": 0.018199114861231748,
      "grad_norm": 1060.744140625,
      "learning_rate": 1.777777777777778e-06,
      "loss": 350.8291,
      "step": 2200
    },
    {
      "epoch": 0.018281838110600985,
      "grad_norm": 2863.481201171875,
      "learning_rate": 1.7858585858585862e-06,
      "loss": 347.886,
      "step": 2210
    },
    {
      "epoch": 0.01836456135997022,
      "grad_norm": 1049.102294921875,
      "learning_rate": 1.793939393939394e-06,
      "loss": 255.8531,
      "step": 2220
    },
    {
      "epoch": 0.018447284609339455,
      "grad_norm": 1447.9163818359375,
      "learning_rate": 1.802020202020202e-06,
      "loss": 270.9402,
      "step": 2230
    },
    {
      "epoch": 0.01853000785870869,
      "grad_norm": 5666.75634765625,
      "learning_rate": 1.8101010101010103e-06,
      "loss": 265.4302,
      "step": 2240
    },
    {
      "epoch": 0.018612731108077925,
      "grad_norm": 901.1162109375,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 327.2734,
      "step": 2250
    },
    {
      "epoch": 0.01869545435744716,
      "grad_norm": 1582.9307861328125,
      "learning_rate": 1.8262626262626265e-06,
      "loss": 295.4947,
      "step": 2260
    },
    {
      "epoch": 0.018778177606816396,
      "grad_norm": 910.0087280273438,
      "learning_rate": 1.8343434343434343e-06,
      "loss": 254.5832,
      "step": 2270
    },
    {
      "epoch": 0.018860900856185633,
      "grad_norm": 1553.7003173828125,
      "learning_rate": 1.8424242424242426e-06,
      "loss": 312.1798,
      "step": 2280
    },
    {
      "epoch": 0.018943624105554866,
      "grad_norm": 1657.673828125,
      "learning_rate": 1.8505050505050506e-06,
      "loss": 333.7471,
      "step": 2290
    },
    {
      "epoch": 0.019026347354924103,
      "grad_norm": 2194.053466796875,
      "learning_rate": 1.8585858585858588e-06,
      "loss": 252.629,
      "step": 2300
    },
    {
      "epoch": 0.019109070604293336,
      "grad_norm": 968.1193237304688,
      "learning_rate": 1.8666666666666669e-06,
      "loss": 317.2808,
      "step": 2310
    },
    {
      "epoch": 0.019191793853662573,
      "grad_norm": 2099.283203125,
      "learning_rate": 1.874747474747475e-06,
      "loss": 318.3145,
      "step": 2320
    },
    {
      "epoch": 0.019274517103031807,
      "grad_norm": 1018.1463012695312,
      "learning_rate": 1.882828282828283e-06,
      "loss": 310.5461,
      "step": 2330
    },
    {
      "epoch": 0.019357240352401044,
      "grad_norm": 1319.893310546875,
      "learning_rate": 1.890909090909091e-06,
      "loss": 403.0323,
      "step": 2340
    },
    {
      "epoch": 0.019439963601770277,
      "grad_norm": 3296.14404296875,
      "learning_rate": 1.8989898989898992e-06,
      "loss": 277.1991,
      "step": 2350
    },
    {
      "epoch": 0.019522686851139514,
      "grad_norm": 2120.728515625,
      "learning_rate": 1.9070707070707072e-06,
      "loss": 266.7476,
      "step": 2360
    },
    {
      "epoch": 0.019605410100508747,
      "grad_norm": 1736.4859619140625,
      "learning_rate": 1.9151515151515154e-06,
      "loss": 385.987,
      "step": 2370
    },
    {
      "epoch": 0.019688133349877984,
      "grad_norm": 1218.7796630859375,
      "learning_rate": 1.9232323232323235e-06,
      "loss": 337.6978,
      "step": 2380
    },
    {
      "epoch": 0.019770856599247218,
      "grad_norm": 4797.71337890625,
      "learning_rate": 1.9313131313131315e-06,
      "loss": 255.6176,
      "step": 2390
    },
    {
      "epoch": 0.019853579848616455,
      "grad_norm": 1956.2137451171875,
      "learning_rate": 1.9393939393939395e-06,
      "loss": 283.0651,
      "step": 2400
    },
    {
      "epoch": 0.019936303097985688,
      "grad_norm": 918.0914916992188,
      "learning_rate": 1.9474747474747475e-06,
      "loss": 312.682,
      "step": 2410
    },
    {
      "epoch": 0.020019026347354925,
      "grad_norm": 2107.04736328125,
      "learning_rate": 1.955555555555556e-06,
      "loss": 384.748,
      "step": 2420
    },
    {
      "epoch": 0.02010174959672416,
      "grad_norm": 861.828125,
      "learning_rate": 1.9636363636363636e-06,
      "loss": 340.8865,
      "step": 2430
    },
    {
      "epoch": 0.020184472846093395,
      "grad_norm": 6480.6171875,
      "learning_rate": 1.9717171717171716e-06,
      "loss": 416.4972,
      "step": 2440
    },
    {
      "epoch": 0.02026719609546263,
      "grad_norm": 1926.79296875,
      "learning_rate": 1.97979797979798e-06,
      "loss": 260.5365,
      "step": 2450
    },
    {
      "epoch": 0.020349919344831865,
      "grad_norm": 1225.735107421875,
      "learning_rate": 1.987878787878788e-06,
      "loss": 261.068,
      "step": 2460
    },
    {
      "epoch": 0.0204326425942011,
      "grad_norm": 857.5597534179688,
      "learning_rate": 1.995959595959596e-06,
      "loss": 241.7558,
      "step": 2470
    },
    {
      "epoch": 0.020515365843570336,
      "grad_norm": 42987.5546875,
      "learning_rate": 2.004040404040404e-06,
      "loss": 380.0271,
      "step": 2480
    },
    {
      "epoch": 0.02059808909293957,
      "grad_norm": 1689.1239013671875,
      "learning_rate": 2.012121212121212e-06,
      "loss": 390.1416,
      "step": 2490
    },
    {
      "epoch": 0.020680812342308806,
      "grad_norm": 4134.138671875,
      "learning_rate": 2.02020202020202e-06,
      "loss": 265.3958,
      "step": 2500
    },
    {
      "epoch": 0.02076353559167804,
      "grad_norm": 1039.45703125,
      "learning_rate": 2.0282828282828286e-06,
      "loss": 297.8323,
      "step": 2510
    },
    {
      "epoch": 0.020846258841047276,
      "grad_norm": 2275.5380859375,
      "learning_rate": 2.0363636363636367e-06,
      "loss": 326.8338,
      "step": 2520
    },
    {
      "epoch": 0.02092898209041651,
      "grad_norm": 1341.4542236328125,
      "learning_rate": 2.0444444444444447e-06,
      "loss": 255.4626,
      "step": 2530
    },
    {
      "epoch": 0.021011705339785747,
      "grad_norm": 1704.446044921875,
      "learning_rate": 2.0525252525252527e-06,
      "loss": 311.6551,
      "step": 2540
    },
    {
      "epoch": 0.021094428589154984,
      "grad_norm": 1272.9852294921875,
      "learning_rate": 2.0606060606060607e-06,
      "loss": 278.1134,
      "step": 2550
    },
    {
      "epoch": 0.021177151838524217,
      "grad_norm": 788.3895263671875,
      "learning_rate": 2.0686868686868688e-06,
      "loss": 282.1122,
      "step": 2560
    },
    {
      "epoch": 0.021259875087893454,
      "grad_norm": 1385.612548828125,
      "learning_rate": 2.0767676767676768e-06,
      "loss": 314.3244,
      "step": 2570
    },
    {
      "epoch": 0.021342598337262687,
      "grad_norm": 1955.3095703125,
      "learning_rate": 2.0848484848484852e-06,
      "loss": 262.7597,
      "step": 2580
    },
    {
      "epoch": 0.021425321586631924,
      "grad_norm": 1727.3680419921875,
      "learning_rate": 2.092929292929293e-06,
      "loss": 264.8611,
      "step": 2590
    },
    {
      "epoch": 0.021508044836001158,
      "grad_norm": 1702.6575927734375,
      "learning_rate": 2.1010101010101013e-06,
      "loss": 286.7952,
      "step": 2600
    },
    {
      "epoch": 0.021590768085370395,
      "grad_norm": 1000.8258666992188,
      "learning_rate": 2.1090909090909093e-06,
      "loss": 259.9386,
      "step": 2610
    },
    {
      "epoch": 0.021673491334739628,
      "grad_norm": 1225.5982666015625,
      "learning_rate": 2.1171717171717173e-06,
      "loss": 248.9738,
      "step": 2620
    },
    {
      "epoch": 0.021756214584108865,
      "grad_norm": 2182.8955078125,
      "learning_rate": 2.1252525252525254e-06,
      "loss": 290.0493,
      "step": 2630
    },
    {
      "epoch": 0.0218389378334781,
      "grad_norm": 946.896484375,
      "learning_rate": 2.133333333333334e-06,
      "loss": 285.7407,
      "step": 2640
    },
    {
      "epoch": 0.021921661082847335,
      "grad_norm": 965.7489624023438,
      "learning_rate": 2.1414141414141414e-06,
      "loss": 277.0386,
      "step": 2650
    },
    {
      "epoch": 0.02200438433221657,
      "grad_norm": 1855.605712890625,
      "learning_rate": 2.1494949494949494e-06,
      "loss": 308.2425,
      "step": 2660
    },
    {
      "epoch": 0.022087107581585805,
      "grad_norm": 945.2372436523438,
      "learning_rate": 2.157575757575758e-06,
      "loss": 314.9084,
      "step": 2670
    },
    {
      "epoch": 0.02216983083095504,
      "grad_norm": 2101.0771484375,
      "learning_rate": 2.165656565656566e-06,
      "loss": 277.5356,
      "step": 2680
    },
    {
      "epoch": 0.022252554080324276,
      "grad_norm": 931.0570678710938,
      "learning_rate": 2.173737373737374e-06,
      "loss": 380.9075,
      "step": 2690
    },
    {
      "epoch": 0.02233527732969351,
      "grad_norm": 849.65380859375,
      "learning_rate": 2.181818181818182e-06,
      "loss": 266.7404,
      "step": 2700
    },
    {
      "epoch": 0.022418000579062746,
      "grad_norm": 795.3492431640625,
      "learning_rate": 2.18989898989899e-06,
      "loss": 301.0031,
      "step": 2710
    },
    {
      "epoch": 0.02250072382843198,
      "grad_norm": 998.2992553710938,
      "learning_rate": 2.197979797979798e-06,
      "loss": 236.0826,
      "step": 2720
    },
    {
      "epoch": 0.022583447077801216,
      "grad_norm": 3029.24462890625,
      "learning_rate": 2.2060606060606064e-06,
      "loss": 284.6466,
      "step": 2730
    },
    {
      "epoch": 0.02266617032717045,
      "grad_norm": 991.4857177734375,
      "learning_rate": 2.2141414141414145e-06,
      "loss": 480.9545,
      "step": 2740
    },
    {
      "epoch": 0.022748893576539687,
      "grad_norm": 610.8831787109375,
      "learning_rate": 2.222222222222222e-06,
      "loss": 280.1348,
      "step": 2750
    },
    {
      "epoch": 0.02283161682590892,
      "grad_norm": 1224.61083984375,
      "learning_rate": 2.2303030303030305e-06,
      "loss": 287.0002,
      "step": 2760
    },
    {
      "epoch": 0.022914340075278157,
      "grad_norm": 967.6873779296875,
      "learning_rate": 2.2383838383838385e-06,
      "loss": 302.4269,
      "step": 2770
    },
    {
      "epoch": 0.02299706332464739,
      "grad_norm": 1249.24169921875,
      "learning_rate": 2.2464646464646466e-06,
      "loss": 324.5018,
      "step": 2780
    },
    {
      "epoch": 0.023079786574016627,
      "grad_norm": 2674.385498046875,
      "learning_rate": 2.254545454545455e-06,
      "loss": 253.8143,
      "step": 2790
    },
    {
      "epoch": 0.023162509823385864,
      "grad_norm": 1356.0936279296875,
      "learning_rate": 2.262626262626263e-06,
      "loss": 287.3596,
      "step": 2800
    },
    {
      "epoch": 0.023245233072755098,
      "grad_norm": 1003.8848266601562,
      "learning_rate": 2.2707070707070706e-06,
      "loss": 259.5809,
      "step": 2810
    },
    {
      "epoch": 0.023327956322124335,
      "grad_norm": 1609.186767578125,
      "learning_rate": 2.278787878787879e-06,
      "loss": 243.4104,
      "step": 2820
    },
    {
      "epoch": 0.023410679571493568,
      "grad_norm": 1286.5640869140625,
      "learning_rate": 2.286868686868687e-06,
      "loss": 303.2196,
      "step": 2830
    },
    {
      "epoch": 0.023493402820862805,
      "grad_norm": 1285.9228515625,
      "learning_rate": 2.294949494949495e-06,
      "loss": 274.7912,
      "step": 2840
    },
    {
      "epoch": 0.02357612607023204,
      "grad_norm": 1015.9469604492188,
      "learning_rate": 2.303030303030303e-06,
      "loss": 255.5114,
      "step": 2850
    },
    {
      "epoch": 0.023658849319601275,
      "grad_norm": 1524.3387451171875,
      "learning_rate": 2.311111111111111e-06,
      "loss": 394.2463,
      "step": 2860
    },
    {
      "epoch": 0.02374157256897051,
      "grad_norm": 4101.37548828125,
      "learning_rate": 2.3191919191919192e-06,
      "loss": 343.4752,
      "step": 2870
    },
    {
      "epoch": 0.023824295818339745,
      "grad_norm": 1837.815673828125,
      "learning_rate": 2.3272727272727277e-06,
      "loss": 306.1534,
      "step": 2880
    },
    {
      "epoch": 0.02390701906770898,
      "grad_norm": 1742.3822021484375,
      "learning_rate": 2.3353535353535357e-06,
      "loss": 382.2382,
      "step": 2890
    },
    {
      "epoch": 0.023989742317078216,
      "grad_norm": 922.0031127929688,
      "learning_rate": 2.3434343434343437e-06,
      "loss": 230.2729,
      "step": 2900
    },
    {
      "epoch": 0.02407246556644745,
      "grad_norm": 860.6764526367188,
      "learning_rate": 2.3515151515151517e-06,
      "loss": 257.8272,
      "step": 2910
    },
    {
      "epoch": 0.024155188815816686,
      "grad_norm": 3094.787841796875,
      "learning_rate": 2.3595959595959598e-06,
      "loss": 247.0388,
      "step": 2920
    },
    {
      "epoch": 0.02423791206518592,
      "grad_norm": 1400.10009765625,
      "learning_rate": 2.367676767676768e-06,
      "loss": 350.3922,
      "step": 2930
    },
    {
      "epoch": 0.024320635314555156,
      "grad_norm": 1306.7431640625,
      "learning_rate": 2.375757575757576e-06,
      "loss": 318.7807,
      "step": 2940
    },
    {
      "epoch": 0.02440335856392439,
      "grad_norm": 859.3858032226562,
      "learning_rate": 2.3838383838383843e-06,
      "loss": 292.0828,
      "step": 2950
    },
    {
      "epoch": 0.024486081813293627,
      "grad_norm": 1207.4444580078125,
      "learning_rate": 2.3919191919191923e-06,
      "loss": 268.6456,
      "step": 2960
    },
    {
      "epoch": 0.02456880506266286,
      "grad_norm": 1388.7242431640625,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 327.8167,
      "step": 2970
    },
    {
      "epoch": 0.024651528312032097,
      "grad_norm": 1601.5413818359375,
      "learning_rate": 2.4080808080808083e-06,
      "loss": 217.4461,
      "step": 2980
    },
    {
      "epoch": 0.02473425156140133,
      "grad_norm": 1094.072509765625,
      "learning_rate": 2.4161616161616164e-06,
      "loss": 237.2505,
      "step": 2990
    },
    {
      "epoch": 0.024816974810770567,
      "grad_norm": 1064.4403076171875,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 274.2577,
      "step": 3000
    },
    {
      "epoch": 0.0248996980601398,
      "grad_norm": 2712.60986328125,
      "learning_rate": 2.432323232323233e-06,
      "loss": 336.2089,
      "step": 3010
    },
    {
      "epoch": 0.024982421309509038,
      "grad_norm": 657.0144653320312,
      "learning_rate": 2.4404040404040404e-06,
      "loss": 285.5267,
      "step": 3020
    },
    {
      "epoch": 0.02506514455887827,
      "grad_norm": 1818.3843994140625,
      "learning_rate": 2.4484848484848485e-06,
      "loss": 345.3333,
      "step": 3030
    },
    {
      "epoch": 0.025147867808247508,
      "grad_norm": 1924.058349609375,
      "learning_rate": 2.456565656565657e-06,
      "loss": 254.806,
      "step": 3040
    },
    {
      "epoch": 0.025230591057616745,
      "grad_norm": 926.4883422851562,
      "learning_rate": 2.464646464646465e-06,
      "loss": 248.8595,
      "step": 3050
    },
    {
      "epoch": 0.025313314306985978,
      "grad_norm": 1217.055419921875,
      "learning_rate": 2.472727272727273e-06,
      "loss": 258.9547,
      "step": 3060
    },
    {
      "epoch": 0.025396037556355215,
      "grad_norm": 837.9513549804688,
      "learning_rate": 2.480808080808081e-06,
      "loss": 239.7319,
      "step": 3070
    },
    {
      "epoch": 0.02547876080572445,
      "grad_norm": 702.333740234375,
      "learning_rate": 2.488888888888889e-06,
      "loss": 290.1943,
      "step": 3080
    },
    {
      "epoch": 0.025561484055093685,
      "grad_norm": 1217.6317138671875,
      "learning_rate": 2.496969696969697e-06,
      "loss": 282.1116,
      "step": 3090
    },
    {
      "epoch": 0.02564420730446292,
      "grad_norm": 1642.226806640625,
      "learning_rate": 2.5050505050505055e-06,
      "loss": 242.3538,
      "step": 3100
    },
    {
      "epoch": 0.025726930553832156,
      "grad_norm": 1036.6883544921875,
      "learning_rate": 2.5131313131313135e-06,
      "loss": 339.3384,
      "step": 3110
    },
    {
      "epoch": 0.02580965380320139,
      "grad_norm": 2547.31396484375,
      "learning_rate": 2.5212121212121215e-06,
      "loss": 268.7113,
      "step": 3120
    },
    {
      "epoch": 0.025892377052570626,
      "grad_norm": 2499.7900390625,
      "learning_rate": 2.5292929292929296e-06,
      "loss": 371.6728,
      "step": 3130
    },
    {
      "epoch": 0.02597510030193986,
      "grad_norm": 835.4087524414062,
      "learning_rate": 2.537373737373738e-06,
      "loss": 331.6685,
      "step": 3140
    },
    {
      "epoch": 0.026057823551309096,
      "grad_norm": 1723.9932861328125,
      "learning_rate": 2.5454545454545456e-06,
      "loss": 241.2475,
      "step": 3150
    },
    {
      "epoch": 0.02614054680067833,
      "grad_norm": 892.8982543945312,
      "learning_rate": 2.5535353535353536e-06,
      "loss": 239.1675,
      "step": 3160
    },
    {
      "epoch": 0.026223270050047567,
      "grad_norm": 1679.0596923828125,
      "learning_rate": 2.5616161616161617e-06,
      "loss": 251.3017,
      "step": 3170
    },
    {
      "epoch": 0.0263059932994168,
      "grad_norm": 1314.4605712890625,
      "learning_rate": 2.5696969696969697e-06,
      "loss": 291.6142,
      "step": 3180
    },
    {
      "epoch": 0.026388716548786037,
      "grad_norm": 1959.02490234375,
      "learning_rate": 2.577777777777778e-06,
      "loss": 335.8238,
      "step": 3190
    },
    {
      "epoch": 0.02647143979815527,
      "grad_norm": 1912.5706787109375,
      "learning_rate": 2.585858585858586e-06,
      "loss": 374.3031,
      "step": 3200
    },
    {
      "epoch": 0.026554163047524507,
      "grad_norm": 3296.021728515625,
      "learning_rate": 2.593939393939394e-06,
      "loss": 258.0628,
      "step": 3210
    },
    {
      "epoch": 0.02663688629689374,
      "grad_norm": 1115.4268798828125,
      "learning_rate": 2.602020202020202e-06,
      "loss": 334.8846,
      "step": 3220
    },
    {
      "epoch": 0.026719609546262978,
      "grad_norm": 973.0841064453125,
      "learning_rate": 2.6101010101010107e-06,
      "loss": 275.7638,
      "step": 3230
    },
    {
      "epoch": 0.02680233279563221,
      "grad_norm": 1405.8780517578125,
      "learning_rate": 2.6181818181818187e-06,
      "loss": 290.556,
      "step": 3240
    },
    {
      "epoch": 0.026885056045001448,
      "grad_norm": 2512.458740234375,
      "learning_rate": 2.6262626262626267e-06,
      "loss": 255.2991,
      "step": 3250
    },
    {
      "epoch": 0.02696777929437068,
      "grad_norm": 1015.4577026367188,
      "learning_rate": 2.6343434343434343e-06,
      "loss": 293.1441,
      "step": 3260
    },
    {
      "epoch": 0.027050502543739918,
      "grad_norm": 1590.4383544921875,
      "learning_rate": 2.6424242424242423e-06,
      "loss": 315.3852,
      "step": 3270
    },
    {
      "epoch": 0.02713322579310915,
      "grad_norm": 1241.1497802734375,
      "learning_rate": 2.6505050505050508e-06,
      "loss": 279.314,
      "step": 3280
    },
    {
      "epoch": 0.02721594904247839,
      "grad_norm": 803.2537841796875,
      "learning_rate": 2.658585858585859e-06,
      "loss": 237.7475,
      "step": 3290
    },
    {
      "epoch": 0.027298672291847625,
      "grad_norm": 1308.884765625,
      "learning_rate": 2.666666666666667e-06,
      "loss": 289.3862,
      "step": 3300
    },
    {
      "epoch": 0.02738139554121686,
      "grad_norm": 1037.373291015625,
      "learning_rate": 2.674747474747475e-06,
      "loss": 278.7418,
      "step": 3310
    },
    {
      "epoch": 0.027464118790586096,
      "grad_norm": 1414.7493896484375,
      "learning_rate": 2.6828282828282833e-06,
      "loss": 388.9876,
      "step": 3320
    },
    {
      "epoch": 0.02754684203995533,
      "grad_norm": 1686.8880615234375,
      "learning_rate": 2.6909090909090913e-06,
      "loss": 224.3872,
      "step": 3330
    },
    {
      "epoch": 0.027629565289324566,
      "grad_norm": 1130.1907958984375,
      "learning_rate": 2.6989898989898994e-06,
      "loss": 219.274,
      "step": 3340
    },
    {
      "epoch": 0.0277122885386938,
      "grad_norm": 1604.05419921875,
      "learning_rate": 2.7070707070707074e-06,
      "loss": 313.1014,
      "step": 3350
    },
    {
      "epoch": 0.027795011788063036,
      "grad_norm": 1515.8199462890625,
      "learning_rate": 2.715151515151516e-06,
      "loss": 351.9699,
      "step": 3360
    },
    {
      "epoch": 0.02787773503743227,
      "grad_norm": 1105.636474609375,
      "learning_rate": 2.7232323232323234e-06,
      "loss": 258.1985,
      "step": 3370
    },
    {
      "epoch": 0.027960458286801507,
      "grad_norm": 2343.5537109375,
      "learning_rate": 2.7313131313131315e-06,
      "loss": 247.2701,
      "step": 3380
    },
    {
      "epoch": 0.02804318153617074,
      "grad_norm": 3364.70068359375,
      "learning_rate": 2.7393939393939395e-06,
      "loss": 320.3825,
      "step": 3390
    },
    {
      "epoch": 0.028125904785539977,
      "grad_norm": 1226.61279296875,
      "learning_rate": 2.7474747474747475e-06,
      "loss": 308.3539,
      "step": 3400
    },
    {
      "epoch": 0.02820862803490921,
      "grad_norm": 1365.4581298828125,
      "learning_rate": 2.755555555555556e-06,
      "loss": 265.8813,
      "step": 3410
    },
    {
      "epoch": 0.028291351284278447,
      "grad_norm": 1471.4437255859375,
      "learning_rate": 2.763636363636364e-06,
      "loss": 246.5465,
      "step": 3420
    },
    {
      "epoch": 0.02837407453364768,
      "grad_norm": 1522.9873046875,
      "learning_rate": 2.771717171717172e-06,
      "loss": 294.6199,
      "step": 3430
    },
    {
      "epoch": 0.028456797783016918,
      "grad_norm": 2295.73095703125,
      "learning_rate": 2.77979797979798e-06,
      "loss": 471.7754,
      "step": 3440
    },
    {
      "epoch": 0.02853952103238615,
      "grad_norm": 1030.5484619140625,
      "learning_rate": 2.7878787878787885e-06,
      "loss": 315.6531,
      "step": 3450
    },
    {
      "epoch": 0.028622244281755388,
      "grad_norm": 1534.0970458984375,
      "learning_rate": 2.7959595959595965e-06,
      "loss": 236.7922,
      "step": 3460
    },
    {
      "epoch": 0.02870496753112462,
      "grad_norm": 1193.41748046875,
      "learning_rate": 2.804040404040404e-06,
      "loss": 301.1876,
      "step": 3470
    },
    {
      "epoch": 0.028787690780493858,
      "grad_norm": 884.4568481445312,
      "learning_rate": 2.812121212121212e-06,
      "loss": 248.0099,
      "step": 3480
    },
    {
      "epoch": 0.02887041402986309,
      "grad_norm": 1417.0870361328125,
      "learning_rate": 2.82020202020202e-06,
      "loss": 214.0175,
      "step": 3490
    },
    {
      "epoch": 0.02895313727923233,
      "grad_norm": 906.0728149414062,
      "learning_rate": 2.8282828282828286e-06,
      "loss": 337.6909,
      "step": 3500
    },
    {
      "epoch": 0.029035860528601562,
      "grad_norm": 720.5711059570312,
      "learning_rate": 2.8363636363636366e-06,
      "loss": 237.9402,
      "step": 3510
    },
    {
      "epoch": 0.0291185837779708,
      "grad_norm": 2159.7626953125,
      "learning_rate": 2.8444444444444446e-06,
      "loss": 290.2204,
      "step": 3520
    },
    {
      "epoch": 0.029201307027340032,
      "grad_norm": 1583.6937255859375,
      "learning_rate": 2.8525252525252527e-06,
      "loss": 339.3829,
      "step": 3530
    },
    {
      "epoch": 0.02928403027670927,
      "grad_norm": 1711.7625732421875,
      "learning_rate": 2.860606060606061e-06,
      "loss": 338.8203,
      "step": 3540
    },
    {
      "epoch": 0.029366753526078506,
      "grad_norm": 1867.3909912109375,
      "learning_rate": 2.868686868686869e-06,
      "loss": 295.4354,
      "step": 3550
    },
    {
      "epoch": 0.02944947677544774,
      "grad_norm": 1350.51708984375,
      "learning_rate": 2.876767676767677e-06,
      "loss": 255.1478,
      "step": 3560
    },
    {
      "epoch": 0.029532200024816976,
      "grad_norm": 1835.9559326171875,
      "learning_rate": 2.884848484848485e-06,
      "loss": 235.7582,
      "step": 3570
    },
    {
      "epoch": 0.02961492327418621,
      "grad_norm": 1273.539794921875,
      "learning_rate": 2.892929292929293e-06,
      "loss": 305.7688,
      "step": 3580
    },
    {
      "epoch": 0.029697646523555447,
      "grad_norm": 1092.1051025390625,
      "learning_rate": 2.9010101010101012e-06,
      "loss": 229.7812,
      "step": 3590
    },
    {
      "epoch": 0.02978036977292468,
      "grad_norm": 1232.7423095703125,
      "learning_rate": 2.9090909090909093e-06,
      "loss": 308.136,
      "step": 3600
    },
    {
      "epoch": 0.029863093022293917,
      "grad_norm": 860.4628295898438,
      "learning_rate": 2.9171717171717173e-06,
      "loss": 245.5094,
      "step": 3610
    },
    {
      "epoch": 0.02994581627166315,
      "grad_norm": 1105.772705078125,
      "learning_rate": 2.9252525252525253e-06,
      "loss": 225.044,
      "step": 3620
    },
    {
      "epoch": 0.030028539521032387,
      "grad_norm": 1092.992431640625,
      "learning_rate": 2.9333333333333338e-06,
      "loss": 240.9013,
      "step": 3630
    },
    {
      "epoch": 0.03011126277040162,
      "grad_norm": 1877.3671875,
      "learning_rate": 2.941414141414142e-06,
      "loss": 304.4779,
      "step": 3640
    },
    {
      "epoch": 0.030193986019770858,
      "grad_norm": 1553.412109375,
      "learning_rate": 2.94949494949495e-06,
      "loss": 253.1373,
      "step": 3650
    },
    {
      "epoch": 0.03027670926914009,
      "grad_norm": 1163.033447265625,
      "learning_rate": 2.957575757575758e-06,
      "loss": 272.4007,
      "step": 3660
    },
    {
      "epoch": 0.030359432518509328,
      "grad_norm": 1105.1661376953125,
      "learning_rate": 2.9656565656565663e-06,
      "loss": 243.5992,
      "step": 3670
    },
    {
      "epoch": 0.03044215576787856,
      "grad_norm": 1370.2781982421875,
      "learning_rate": 2.9737373737373743e-06,
      "loss": 232.3612,
      "step": 3680
    },
    {
      "epoch": 0.030524879017247798,
      "grad_norm": 925.8299560546875,
      "learning_rate": 2.981818181818182e-06,
      "loss": 194.9076,
      "step": 3690
    },
    {
      "epoch": 0.03060760226661703,
      "grad_norm": 4565.89208984375,
      "learning_rate": 2.98989898989899e-06,
      "loss": 317.9731,
      "step": 3700
    },
    {
      "epoch": 0.03069032551598627,
      "grad_norm": 1785.587890625,
      "learning_rate": 2.997979797979798e-06,
      "loss": 261.7598,
      "step": 3710
    },
    {
      "epoch": 0.030773048765355502,
      "grad_norm": 1177.1761474609375,
      "learning_rate": 3.0060606060606064e-06,
      "loss": 211.625,
      "step": 3720
    },
    {
      "epoch": 0.03085577201472474,
      "grad_norm": 980.7220458984375,
      "learning_rate": 3.0141414141414144e-06,
      "loss": 198.16,
      "step": 3730
    },
    {
      "epoch": 0.030938495264093972,
      "grad_norm": 1450.34033203125,
      "learning_rate": 3.0222222222222225e-06,
      "loss": 281.4796,
      "step": 3740
    },
    {
      "epoch": 0.03102121851346321,
      "grad_norm": 917.07275390625,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 266.2501,
      "step": 3750
    },
    {
      "epoch": 0.031103941762832443,
      "grad_norm": 1657.814697265625,
      "learning_rate": 3.038383838383839e-06,
      "loss": 279.4682,
      "step": 3760
    },
    {
      "epoch": 0.03118666501220168,
      "grad_norm": 1294.77685546875,
      "learning_rate": 3.046464646464647e-06,
      "loss": 259.5207,
      "step": 3770
    },
    {
      "epoch": 0.03126938826157091,
      "grad_norm": 1622.716796875,
      "learning_rate": 3.054545454545455e-06,
      "loss": 298.1313,
      "step": 3780
    },
    {
      "epoch": 0.031352111510940146,
      "grad_norm": 1660.8116455078125,
      "learning_rate": 3.0626262626262626e-06,
      "loss": 250.9933,
      "step": 3790
    },
    {
      "epoch": 0.03143483476030939,
      "grad_norm": 799.0031127929688,
      "learning_rate": 3.0707070707070706e-06,
      "loss": 310.6956,
      "step": 3800
    },
    {
      "epoch": 0.03151755800967862,
      "grad_norm": 1199.9678955078125,
      "learning_rate": 3.078787878787879e-06,
      "loss": 283.1158,
      "step": 3810
    },
    {
      "epoch": 0.031600281259047854,
      "grad_norm": 1449.6212158203125,
      "learning_rate": 3.086868686868687e-06,
      "loss": 249.93,
      "step": 3820
    },
    {
      "epoch": 0.031683004508417094,
      "grad_norm": 1299.2960205078125,
      "learning_rate": 3.094949494949495e-06,
      "loss": 224.7813,
      "step": 3830
    },
    {
      "epoch": 0.03176572775778633,
      "grad_norm": 2239.721435546875,
      "learning_rate": 3.103030303030303e-06,
      "loss": 223.2453,
      "step": 3840
    },
    {
      "epoch": 0.03184845100715556,
      "grad_norm": 1760.5396728515625,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 252.7971,
      "step": 3850
    },
    {
      "epoch": 0.031931174256524794,
      "grad_norm": 682.6892700195312,
      "learning_rate": 3.1191919191919196e-06,
      "loss": 216.5466,
      "step": 3860
    },
    {
      "epoch": 0.032013897505894034,
      "grad_norm": 2052.32861328125,
      "learning_rate": 3.1272727272727276e-06,
      "loss": 288.775,
      "step": 3870
    },
    {
      "epoch": 0.03209662075526327,
      "grad_norm": 1602.2225341796875,
      "learning_rate": 3.1353535353535357e-06,
      "loss": 328.757,
      "step": 3880
    },
    {
      "epoch": 0.0321793440046325,
      "grad_norm": 681.2459106445312,
      "learning_rate": 3.143434343434344e-06,
      "loss": 190.3252,
      "step": 3890
    },
    {
      "epoch": 0.032262067254001735,
      "grad_norm": 1182.54052734375,
      "learning_rate": 3.1515151515151517e-06,
      "loss": 268.2503,
      "step": 3900
    },
    {
      "epoch": 0.032344790503370975,
      "grad_norm": 1219.3660888671875,
      "learning_rate": 3.1595959595959597e-06,
      "loss": 239.1843,
      "step": 3910
    },
    {
      "epoch": 0.03242751375274021,
      "grad_norm": 1041.707275390625,
      "learning_rate": 3.1676767676767678e-06,
      "loss": 273.016,
      "step": 3920
    },
    {
      "epoch": 0.03251023700210944,
      "grad_norm": 1269.7681884765625,
      "learning_rate": 3.1757575757575758e-06,
      "loss": 249.6547,
      "step": 3930
    },
    {
      "epoch": 0.032592960251478675,
      "grad_norm": 1372.903564453125,
      "learning_rate": 3.1838383838383842e-06,
      "loss": 311.8126,
      "step": 3940
    },
    {
      "epoch": 0.032675683500847916,
      "grad_norm": 2617.830322265625,
      "learning_rate": 3.1919191919191923e-06,
      "loss": 299.153,
      "step": 3950
    },
    {
      "epoch": 0.03275840675021715,
      "grad_norm": 1350.3302001953125,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 238.4413,
      "step": 3960
    },
    {
      "epoch": 0.03284112999958638,
      "grad_norm": 2270.457763671875,
      "learning_rate": 3.2080808080808083e-06,
      "loss": 267.5316,
      "step": 3970
    },
    {
      "epoch": 0.032923853248955616,
      "grad_norm": 1409.05712890625,
      "learning_rate": 3.2161616161616168e-06,
      "loss": 237.9799,
      "step": 3980
    },
    {
      "epoch": 0.033006576498324856,
      "grad_norm": 1641.050537109375,
      "learning_rate": 3.2242424242424248e-06,
      "loss": 352.6279,
      "step": 3990
    },
    {
      "epoch": 0.03308929974769409,
      "grad_norm": 1178.6591796875,
      "learning_rate": 3.232323232323233e-06,
      "loss": 200.0928,
      "step": 4000
    },
    {
      "epoch": 0.03317202299706332,
      "grad_norm": 895.1983032226562,
      "learning_rate": 3.2404040404040404e-06,
      "loss": 274.1366,
      "step": 4010
    },
    {
      "epoch": 0.03325474624643256,
      "grad_norm": 2193.698486328125,
      "learning_rate": 3.2484848484848484e-06,
      "loss": 205.6771,
      "step": 4020
    },
    {
      "epoch": 0.0333374694958018,
      "grad_norm": 1705.0743408203125,
      "learning_rate": 3.256565656565657e-06,
      "loss": 253.3518,
      "step": 4030
    },
    {
      "epoch": 0.03342019274517103,
      "grad_norm": 1552.232666015625,
      "learning_rate": 3.264646464646465e-06,
      "loss": 243.4265,
      "step": 4040
    },
    {
      "epoch": 0.033502915994540264,
      "grad_norm": 4037.46875,
      "learning_rate": 3.272727272727273e-06,
      "loss": 229.4004,
      "step": 4050
    },
    {
      "epoch": 0.033585639243909504,
      "grad_norm": 3386.515869140625,
      "learning_rate": 3.280808080808081e-06,
      "loss": 338.8573,
      "step": 4060
    },
    {
      "epoch": 0.03366836249327874,
      "grad_norm": 1145.6063232421875,
      "learning_rate": 3.2888888888888894e-06,
      "loss": 254.9263,
      "step": 4070
    },
    {
      "epoch": 0.03375108574264797,
      "grad_norm": 2452.519775390625,
      "learning_rate": 3.2969696969696974e-06,
      "loss": 365.3762,
      "step": 4080
    },
    {
      "epoch": 0.033833808992017204,
      "grad_norm": 1752.8304443359375,
      "learning_rate": 3.3050505050505054e-06,
      "loss": 242.0698,
      "step": 4090
    },
    {
      "epoch": 0.033916532241386445,
      "grad_norm": 1975.453857421875,
      "learning_rate": 3.3131313131313135e-06,
      "loss": 294.2446,
      "step": 4100
    },
    {
      "epoch": 0.03399925549075568,
      "grad_norm": 1159.05615234375,
      "learning_rate": 3.321212121212121e-06,
      "loss": 254.9454,
      "step": 4110
    },
    {
      "epoch": 0.03408197874012491,
      "grad_norm": 904.747802734375,
      "learning_rate": 3.3292929292929295e-06,
      "loss": 291.38,
      "step": 4120
    },
    {
      "epoch": 0.034164701989494145,
      "grad_norm": 1563.5689697265625,
      "learning_rate": 3.3373737373737375e-06,
      "loss": 212.5943,
      "step": 4130
    },
    {
      "epoch": 0.034247425238863385,
      "grad_norm": 1204.7821044921875,
      "learning_rate": 3.3454545454545456e-06,
      "loss": 265.8186,
      "step": 4140
    },
    {
      "epoch": 0.03433014848823262,
      "grad_norm": 904.9209594726562,
      "learning_rate": 3.3535353535353536e-06,
      "loss": 260.1104,
      "step": 4150
    },
    {
      "epoch": 0.03441287173760185,
      "grad_norm": 1248.727783203125,
      "learning_rate": 3.361616161616162e-06,
      "loss": 291.2336,
      "step": 4160
    },
    {
      "epoch": 0.034495594986971086,
      "grad_norm": 893.3284912109375,
      "learning_rate": 3.36969696969697e-06,
      "loss": 233.5644,
      "step": 4170
    },
    {
      "epoch": 0.034578318236340326,
      "grad_norm": 1158.58984375,
      "learning_rate": 3.377777777777778e-06,
      "loss": 246.782,
      "step": 4180
    },
    {
      "epoch": 0.03466104148570956,
      "grad_norm": 1969.4088134765625,
      "learning_rate": 3.385858585858586e-06,
      "loss": 237.9205,
      "step": 4190
    },
    {
      "epoch": 0.03474376473507879,
      "grad_norm": 1035.88427734375,
      "learning_rate": 3.3939393939393946e-06,
      "loss": 321.7581,
      "step": 4200
    },
    {
      "epoch": 0.034826487984448026,
      "grad_norm": 918.1465454101562,
      "learning_rate": 3.4020202020202026e-06,
      "loss": 272.7782,
      "step": 4210
    },
    {
      "epoch": 0.03490921123381727,
      "grad_norm": 1070.0615234375,
      "learning_rate": 3.41010101010101e-06,
      "loss": 224.1239,
      "step": 4220
    },
    {
      "epoch": 0.0349919344831865,
      "grad_norm": 1230.9432373046875,
      "learning_rate": 3.4181818181818182e-06,
      "loss": 276.4893,
      "step": 4230
    },
    {
      "epoch": 0.035074657732555733,
      "grad_norm": 1131.748046875,
      "learning_rate": 3.4262626262626262e-06,
      "loss": 217.2287,
      "step": 4240
    },
    {
      "epoch": 0.03515738098192497,
      "grad_norm": 2427.749267578125,
      "learning_rate": 3.4343434343434347e-06,
      "loss": 251.6459,
      "step": 4250
    },
    {
      "epoch": 0.03524010423129421,
      "grad_norm": 1668.80810546875,
      "learning_rate": 3.4424242424242427e-06,
      "loss": 317.6187,
      "step": 4260
    },
    {
      "epoch": 0.03532282748066344,
      "grad_norm": 372.29510498046875,
      "learning_rate": 3.4505050505050507e-06,
      "loss": 179.444,
      "step": 4270
    },
    {
      "epoch": 0.035405550730032674,
      "grad_norm": 1011.8233032226562,
      "learning_rate": 3.4585858585858588e-06,
      "loss": 219.8903,
      "step": 4280
    },
    {
      "epoch": 0.03548827397940191,
      "grad_norm": 1418.88232421875,
      "learning_rate": 3.4666666666666672e-06,
      "loss": 286.1539,
      "step": 4290
    },
    {
      "epoch": 0.03557099722877115,
      "grad_norm": 2228.23095703125,
      "learning_rate": 3.4747474747474752e-06,
      "loss": 293.0923,
      "step": 4300
    },
    {
      "epoch": 0.03565372047814038,
      "grad_norm": 715.6218872070312,
      "learning_rate": 3.4828282828282833e-06,
      "loss": 282.1017,
      "step": 4310
    },
    {
      "epoch": 0.035736443727509615,
      "grad_norm": 742.7416381835938,
      "learning_rate": 3.4909090909090913e-06,
      "loss": 204.9258,
      "step": 4320
    },
    {
      "epoch": 0.035819166976878855,
      "grad_norm": 1228.9534912109375,
      "learning_rate": 3.498989898989899e-06,
      "loss": 259.2133,
      "step": 4330
    },
    {
      "epoch": 0.03590189022624809,
      "grad_norm": 967.5274047851562,
      "learning_rate": 3.5070707070707073e-06,
      "loss": 288.3385,
      "step": 4340
    },
    {
      "epoch": 0.03598461347561732,
      "grad_norm": 1587.67724609375,
      "learning_rate": 3.5151515151515154e-06,
      "loss": 277.6203,
      "step": 4350
    },
    {
      "epoch": 0.036067336724986555,
      "grad_norm": 1431.6448974609375,
      "learning_rate": 3.5232323232323234e-06,
      "loss": 213.064,
      "step": 4360
    },
    {
      "epoch": 0.036150059974355796,
      "grad_norm": 1135.779052734375,
      "learning_rate": 3.5313131313131314e-06,
      "loss": 221.4357,
      "step": 4370
    },
    {
      "epoch": 0.03623278322372503,
      "grad_norm": 1519.3890380859375,
      "learning_rate": 3.53939393939394e-06,
      "loss": 277.6575,
      "step": 4380
    },
    {
      "epoch": 0.03631550647309426,
      "grad_norm": 1360.25048828125,
      "learning_rate": 3.547474747474748e-06,
      "loss": 195.2055,
      "step": 4390
    },
    {
      "epoch": 0.036398229722463496,
      "grad_norm": 2059.44287109375,
      "learning_rate": 3.555555555555556e-06,
      "loss": 324.8405,
      "step": 4400
    },
    {
      "epoch": 0.036480952971832736,
      "grad_norm": 780.7919921875,
      "learning_rate": 3.563636363636364e-06,
      "loss": 212.4319,
      "step": 4410
    },
    {
      "epoch": 0.03656367622120197,
      "grad_norm": 1413.5367431640625,
      "learning_rate": 3.5717171717171724e-06,
      "loss": 255.6041,
      "step": 4420
    },
    {
      "epoch": 0.0366463994705712,
      "grad_norm": 637.59521484375,
      "learning_rate": 3.57979797979798e-06,
      "loss": 193.1592,
      "step": 4430
    },
    {
      "epoch": 0.03672912271994044,
      "grad_norm": 1240.098876953125,
      "learning_rate": 3.587878787878788e-06,
      "loss": 331.8386,
      "step": 4440
    },
    {
      "epoch": 0.03681184596930968,
      "grad_norm": 980.0120239257812,
      "learning_rate": 3.595959595959596e-06,
      "loss": 276.4505,
      "step": 4450
    },
    {
      "epoch": 0.03689456921867891,
      "grad_norm": 2805.927001953125,
      "learning_rate": 3.604040404040404e-06,
      "loss": 269.0381,
      "step": 4460
    },
    {
      "epoch": 0.036977292468048144,
      "grad_norm": 2359.113525390625,
      "learning_rate": 3.6121212121212125e-06,
      "loss": 258.4454,
      "step": 4470
    },
    {
      "epoch": 0.03706001571741738,
      "grad_norm": 1293.63720703125,
      "learning_rate": 3.6202020202020205e-06,
      "loss": 225.2379,
      "step": 4480
    },
    {
      "epoch": 0.03714273896678662,
      "grad_norm": 2243.848876953125,
      "learning_rate": 3.6282828282828286e-06,
      "loss": 213.6965,
      "step": 4490
    },
    {
      "epoch": 0.03722546221615585,
      "grad_norm": 770.7854614257812,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 231.6815,
      "step": 4500
    },
    {
      "epoch": 0.037308185465525084,
      "grad_norm": 1308.199462890625,
      "learning_rate": 3.644444444444445e-06,
      "loss": 232.9941,
      "step": 4510
    },
    {
      "epoch": 0.03739090871489432,
      "grad_norm": 883.5875244140625,
      "learning_rate": 3.652525252525253e-06,
      "loss": 216.6295,
      "step": 4520
    },
    {
      "epoch": 0.03747363196426356,
      "grad_norm": 672.0443115234375,
      "learning_rate": 3.660606060606061e-06,
      "loss": 201.8103,
      "step": 4530
    },
    {
      "epoch": 0.03755635521363279,
      "grad_norm": 1314.2528076171875,
      "learning_rate": 3.6686868686868687e-06,
      "loss": 256.3278,
      "step": 4540
    },
    {
      "epoch": 0.037639078463002025,
      "grad_norm": 1359.0018310546875,
      "learning_rate": 3.6767676767676767e-06,
      "loss": 216.419,
      "step": 4550
    },
    {
      "epoch": 0.037721801712371265,
      "grad_norm": 1884.207763671875,
      "learning_rate": 3.684848484848485e-06,
      "loss": 211.731,
      "step": 4560
    },
    {
      "epoch": 0.0378045249617405,
      "grad_norm": 1213.398193359375,
      "learning_rate": 3.692929292929293e-06,
      "loss": 272.2958,
      "step": 4570
    },
    {
      "epoch": 0.03788724821110973,
      "grad_norm": 2207.1171875,
      "learning_rate": 3.701010101010101e-06,
      "loss": 265.4331,
      "step": 4580
    },
    {
      "epoch": 0.037969971460478966,
      "grad_norm": 1944.73876953125,
      "learning_rate": 3.7090909090909092e-06,
      "loss": 272.1659,
      "step": 4590
    },
    {
      "epoch": 0.038052694709848206,
      "grad_norm": 1226.4189453125,
      "learning_rate": 3.7171717171717177e-06,
      "loss": 226.3827,
      "step": 4600
    },
    {
      "epoch": 0.03813541795921744,
      "grad_norm": 1100.8829345703125,
      "learning_rate": 3.7252525252525257e-06,
      "loss": 221.5738,
      "step": 4610
    },
    {
      "epoch": 0.03821814120858667,
      "grad_norm": 1859.399169921875,
      "learning_rate": 3.7333333333333337e-06,
      "loss": 229.0128,
      "step": 4620
    },
    {
      "epoch": 0.038300864457955906,
      "grad_norm": 1107.61572265625,
      "learning_rate": 3.7414141414141418e-06,
      "loss": 224.0653,
      "step": 4630
    },
    {
      "epoch": 0.03838358770732515,
      "grad_norm": 1617.0789794921875,
      "learning_rate": 3.74949494949495e-06,
      "loss": 240.6666,
      "step": 4640
    },
    {
      "epoch": 0.03846631095669438,
      "grad_norm": 2554.2158203125,
      "learning_rate": 3.757575757575758e-06,
      "loss": 258.9923,
      "step": 4650
    },
    {
      "epoch": 0.038549034206063613,
      "grad_norm": 3580.094482421875,
      "learning_rate": 3.765656565656566e-06,
      "loss": 270.2465,
      "step": 4660
    },
    {
      "epoch": 0.03863175745543285,
      "grad_norm": 1391.0946044921875,
      "learning_rate": 3.773737373737374e-06,
      "loss": 184.8681,
      "step": 4670
    },
    {
      "epoch": 0.03871448070480209,
      "grad_norm": 1413.2205810546875,
      "learning_rate": 3.781818181818182e-06,
      "loss": 210.8876,
      "step": 4680
    },
    {
      "epoch": 0.03879720395417132,
      "grad_norm": 1123.7315673828125,
      "learning_rate": 3.7898989898989903e-06,
      "loss": 226.6469,
      "step": 4690
    },
    {
      "epoch": 0.038879927203540554,
      "grad_norm": 1507.7408447265625,
      "learning_rate": 3.7979797979797984e-06,
      "loss": 243.835,
      "step": 4700
    },
    {
      "epoch": 0.03896265045290979,
      "grad_norm": 1154.9927978515625,
      "learning_rate": 3.8060606060606064e-06,
      "loss": 275.5142,
      "step": 4710
    },
    {
      "epoch": 0.03904537370227903,
      "grad_norm": 1263.9693603515625,
      "learning_rate": 3.8141414141414144e-06,
      "loss": 208.4059,
      "step": 4720
    },
    {
      "epoch": 0.03912809695164826,
      "grad_norm": 2182.939208984375,
      "learning_rate": 3.8222222222222224e-06,
      "loss": 225.735,
      "step": 4730
    },
    {
      "epoch": 0.039210820201017495,
      "grad_norm": 1238.331298828125,
      "learning_rate": 3.830303030303031e-06,
      "loss": 215.773,
      "step": 4740
    },
    {
      "epoch": 0.03929354345038673,
      "grad_norm": 1922.897216796875,
      "learning_rate": 3.8383838383838385e-06,
      "loss": 205.833,
      "step": 4750
    },
    {
      "epoch": 0.03937626669975597,
      "grad_norm": 1248.5205078125,
      "learning_rate": 3.846464646464647e-06,
      "loss": 212.8604,
      "step": 4760
    },
    {
      "epoch": 0.0394589899491252,
      "grad_norm": 1999.810791015625,
      "learning_rate": 3.8545454545454545e-06,
      "loss": 234.7124,
      "step": 4770
    },
    {
      "epoch": 0.039541713198494435,
      "grad_norm": 1900.56787109375,
      "learning_rate": 3.862626262626263e-06,
      "loss": 206.3176,
      "step": 4780
    },
    {
      "epoch": 0.03962443644786367,
      "grad_norm": 1776.326171875,
      "learning_rate": 3.8707070707070706e-06,
      "loss": 235.4771,
      "step": 4790
    },
    {
      "epoch": 0.03970715969723291,
      "grad_norm": 1206.546630859375,
      "learning_rate": 3.878787878787879e-06,
      "loss": 232.6159,
      "step": 4800
    },
    {
      "epoch": 0.03978988294660214,
      "grad_norm": 1193.0535888671875,
      "learning_rate": 3.8868686868686875e-06,
      "loss": 239.4891,
      "step": 4810
    },
    {
      "epoch": 0.039872606195971376,
      "grad_norm": 1266.81103515625,
      "learning_rate": 3.894949494949495e-06,
      "loss": 264.1928,
      "step": 4820
    },
    {
      "epoch": 0.039955329445340616,
      "grad_norm": 1152.444580078125,
      "learning_rate": 3.9030303030303035e-06,
      "loss": 197.2479,
      "step": 4830
    },
    {
      "epoch": 0.04003805269470985,
      "grad_norm": 1796.5777587890625,
      "learning_rate": 3.911111111111112e-06,
      "loss": 232.1052,
      "step": 4840
    },
    {
      "epoch": 0.04012077594407908,
      "grad_norm": 1190.642822265625,
      "learning_rate": 3.9191919191919196e-06,
      "loss": 246.6287,
      "step": 4850
    },
    {
      "epoch": 0.04020349919344832,
      "grad_norm": 1134.314453125,
      "learning_rate": 3.927272727272727e-06,
      "loss": 268.4593,
      "step": 4860
    },
    {
      "epoch": 0.04028622244281756,
      "grad_norm": 1696.9732666015625,
      "learning_rate": 3.935353535353536e-06,
      "loss": 226.5233,
      "step": 4870
    },
    {
      "epoch": 0.04036894569218679,
      "grad_norm": 1237.491455078125,
      "learning_rate": 3.943434343434343e-06,
      "loss": 196.7495,
      "step": 4880
    },
    {
      "epoch": 0.040451668941556024,
      "grad_norm": 2431.97265625,
      "learning_rate": 3.951515151515152e-06,
      "loss": 220.9505,
      "step": 4890
    },
    {
      "epoch": 0.04053439219092526,
      "grad_norm": 1421.2952880859375,
      "learning_rate": 3.95959595959596e-06,
      "loss": 253.9248,
      "step": 4900
    },
    {
      "epoch": 0.0406171154402945,
      "grad_norm": 1305.2431640625,
      "learning_rate": 3.967676767676768e-06,
      "loss": 302.1668,
      "step": 4910
    },
    {
      "epoch": 0.04069983868966373,
      "grad_norm": 1975.3135986328125,
      "learning_rate": 3.975757575757576e-06,
      "loss": 282.536,
      "step": 4920
    },
    {
      "epoch": 0.040782561939032964,
      "grad_norm": 1655.1588134765625,
      "learning_rate": 3.983838383838385e-06,
      "loss": 217.2957,
      "step": 4930
    },
    {
      "epoch": 0.0408652851884022,
      "grad_norm": 1051.86083984375,
      "learning_rate": 3.991919191919192e-06,
      "loss": 240.2383,
      "step": 4940
    },
    {
      "epoch": 0.04094800843777144,
      "grad_norm": 773.5546875,
      "learning_rate": 4.000000000000001e-06,
      "loss": 221.8524,
      "step": 4950
    },
    {
      "epoch": 0.04103073168714067,
      "grad_norm": 1287.1944580078125,
      "learning_rate": 4.008080808080808e-06,
      "loss": 229.3614,
      "step": 4960
    },
    {
      "epoch": 0.041113454936509905,
      "grad_norm": 1897.0220947265625,
      "learning_rate": 4.016161616161616e-06,
      "loss": 233.9596,
      "step": 4970
    },
    {
      "epoch": 0.04119617818587914,
      "grad_norm": 1463.894287109375,
      "learning_rate": 4.024242424242424e-06,
      "loss": 205.995,
      "step": 4980
    },
    {
      "epoch": 0.04127890143524838,
      "grad_norm": 981.1106567382812,
      "learning_rate": 4.032323232323233e-06,
      "loss": 226.2633,
      "step": 4990
    },
    {
      "epoch": 0.04136162468461761,
      "grad_norm": 3404.24365234375,
      "learning_rate": 4.04040404040404e-06,
      "loss": 249.0778,
      "step": 5000
    },
    {
      "epoch": 0.041444347933986846,
      "grad_norm": 1160.4622802734375,
      "learning_rate": 4.048484848484849e-06,
      "loss": 201.7348,
      "step": 5010
    },
    {
      "epoch": 0.04152707118335608,
      "grad_norm": 1339.532470703125,
      "learning_rate": 4.056565656565657e-06,
      "loss": 296.6651,
      "step": 5020
    },
    {
      "epoch": 0.04160979443272532,
      "grad_norm": 1330.253173828125,
      "learning_rate": 4.064646464646465e-06,
      "loss": 222.7661,
      "step": 5030
    },
    {
      "epoch": 0.04169251768209455,
      "grad_norm": 1871.73095703125,
      "learning_rate": 4.072727272727273e-06,
      "loss": 185.6875,
      "step": 5040
    },
    {
      "epoch": 0.041775240931463786,
      "grad_norm": 989.5170288085938,
      "learning_rate": 4.080808080808081e-06,
      "loss": 184.853,
      "step": 5050
    },
    {
      "epoch": 0.04185796418083302,
      "grad_norm": 1325.712158203125,
      "learning_rate": 4.088888888888889e-06,
      "loss": 177.4656,
      "step": 5060
    },
    {
      "epoch": 0.04194068743020226,
      "grad_norm": 1084.450927734375,
      "learning_rate": 4.096969696969697e-06,
      "loss": 218.0059,
      "step": 5070
    },
    {
      "epoch": 0.04202341067957149,
      "grad_norm": 1230.160888671875,
      "learning_rate": 4.105050505050505e-06,
      "loss": 202.8811,
      "step": 5080
    },
    {
      "epoch": 0.04210613392894073,
      "grad_norm": 909.71728515625,
      "learning_rate": 4.113131313131313e-06,
      "loss": 229.2698,
      "step": 5090
    },
    {
      "epoch": 0.04218885717830997,
      "grad_norm": 2449.3154296875,
      "learning_rate": 4.1212121212121215e-06,
      "loss": 211.7621,
      "step": 5100
    },
    {
      "epoch": 0.0422715804276792,
      "grad_norm": 578.8762817382812,
      "learning_rate": 4.12929292929293e-06,
      "loss": 220.8765,
      "step": 5110
    },
    {
      "epoch": 0.042354303677048434,
      "grad_norm": 1294.6402587890625,
      "learning_rate": 4.1373737373737375e-06,
      "loss": 252.3842,
      "step": 5120
    },
    {
      "epoch": 0.04243702692641767,
      "grad_norm": 911.5681762695312,
      "learning_rate": 4.145454545454546e-06,
      "loss": 294.7006,
      "step": 5130
    },
    {
      "epoch": 0.04251975017578691,
      "grad_norm": 1328.2413330078125,
      "learning_rate": 4.1535353535353536e-06,
      "loss": 211.482,
      "step": 5140
    },
    {
      "epoch": 0.04260247342515614,
      "grad_norm": 1352.004638671875,
      "learning_rate": 4.161616161616162e-06,
      "loss": 210.5752,
      "step": 5150
    },
    {
      "epoch": 0.042685196674525375,
      "grad_norm": 1081.876953125,
      "learning_rate": 4.1696969696969705e-06,
      "loss": 224.4739,
      "step": 5160
    },
    {
      "epoch": 0.04276791992389461,
      "grad_norm": 2295.4033203125,
      "learning_rate": 4.177777777777778e-06,
      "loss": 229.6548,
      "step": 5170
    },
    {
      "epoch": 0.04285064317326385,
      "grad_norm": 1036.4749755859375,
      "learning_rate": 4.185858585858586e-06,
      "loss": 208.1364,
      "step": 5180
    },
    {
      "epoch": 0.04293336642263308,
      "grad_norm": 1558.9718017578125,
      "learning_rate": 4.193939393939394e-06,
      "loss": 262.7995,
      "step": 5190
    },
    {
      "epoch": 0.043016089672002315,
      "grad_norm": 1155.65478515625,
      "learning_rate": 4.2020202020202026e-06,
      "loss": 251.0647,
      "step": 5200
    },
    {
      "epoch": 0.04309881292137155,
      "grad_norm": 1081.0423583984375,
      "learning_rate": 4.21010101010101e-06,
      "loss": 235.0402,
      "step": 5210
    },
    {
      "epoch": 0.04318153617074079,
      "grad_norm": 725.3358764648438,
      "learning_rate": 4.218181818181819e-06,
      "loss": 310.5599,
      "step": 5220
    },
    {
      "epoch": 0.04326425942011002,
      "grad_norm": 1182.2801513671875,
      "learning_rate": 4.226262626262626e-06,
      "loss": 178.3826,
      "step": 5230
    },
    {
      "epoch": 0.043346982669479256,
      "grad_norm": 1413.02587890625,
      "learning_rate": 4.234343434343435e-06,
      "loss": 257.1767,
      "step": 5240
    },
    {
      "epoch": 0.04342970591884849,
      "grad_norm": 1222.1142578125,
      "learning_rate": 4.242424242424243e-06,
      "loss": 218.1251,
      "step": 5250
    },
    {
      "epoch": 0.04351242916821773,
      "grad_norm": 998.1055297851562,
      "learning_rate": 4.250505050505051e-06,
      "loss": 211.1936,
      "step": 5260
    },
    {
      "epoch": 0.04359515241758696,
      "grad_norm": 844.65478515625,
      "learning_rate": 4.258585858585859e-06,
      "loss": 235.3665,
      "step": 5270
    },
    {
      "epoch": 0.0436778756669562,
      "grad_norm": 691.6785278320312,
      "learning_rate": 4.266666666666668e-06,
      "loss": 173.3902,
      "step": 5280
    },
    {
      "epoch": 0.04376059891632543,
      "grad_norm": 1113.0621337890625,
      "learning_rate": 4.274747474747475e-06,
      "loss": 169.7838,
      "step": 5290
    },
    {
      "epoch": 0.04384332216569467,
      "grad_norm": 1622.1474609375,
      "learning_rate": 4.282828282828283e-06,
      "loss": 217.8731,
      "step": 5300
    },
    {
      "epoch": 0.043926045415063904,
      "grad_norm": 1753.8427734375,
      "learning_rate": 4.290909090909091e-06,
      "loss": 251.789,
      "step": 5310
    },
    {
      "epoch": 0.04400876866443314,
      "grad_norm": 918.1004638671875,
      "learning_rate": 4.298989898989899e-06,
      "loss": 201.8801,
      "step": 5320
    },
    {
      "epoch": 0.04409149191380238,
      "grad_norm": 899.498779296875,
      "learning_rate": 4.307070707070707e-06,
      "loss": 230.2721,
      "step": 5330
    },
    {
      "epoch": 0.04417421516317161,
      "grad_norm": 1194.7314453125,
      "learning_rate": 4.315151515151516e-06,
      "loss": 302.4559,
      "step": 5340
    },
    {
      "epoch": 0.044256938412540844,
      "grad_norm": 745.4715576171875,
      "learning_rate": 4.323232323232323e-06,
      "loss": 217.5794,
      "step": 5350
    },
    {
      "epoch": 0.04433966166191008,
      "grad_norm": 3024.5693359375,
      "learning_rate": 4.331313131313132e-06,
      "loss": 215.8062,
      "step": 5360
    },
    {
      "epoch": 0.04442238491127932,
      "grad_norm": 692.5128173828125,
      "learning_rate": 4.33939393939394e-06,
      "loss": 206.9033,
      "step": 5370
    },
    {
      "epoch": 0.04450510816064855,
      "grad_norm": 638.6357421875,
      "learning_rate": 4.347474747474748e-06,
      "loss": 277.6233,
      "step": 5380
    },
    {
      "epoch": 0.044587831410017785,
      "grad_norm": 1237.0350341796875,
      "learning_rate": 4.3555555555555555e-06,
      "loss": 202.0422,
      "step": 5390
    },
    {
      "epoch": 0.04467055465938702,
      "grad_norm": 1546.1956787109375,
      "learning_rate": 4.363636363636364e-06,
      "loss": 195.5879,
      "step": 5400
    },
    {
      "epoch": 0.04475327790875626,
      "grad_norm": 1744.332763671875,
      "learning_rate": 4.3717171717171715e-06,
      "loss": 308.2887,
      "step": 5410
    },
    {
      "epoch": 0.04483600115812549,
      "grad_norm": 1885.436767578125,
      "learning_rate": 4.37979797979798e-06,
      "loss": 257.4365,
      "step": 5420
    },
    {
      "epoch": 0.044918724407494726,
      "grad_norm": 1949.8758544921875,
      "learning_rate": 4.387878787878788e-06,
      "loss": 288.0396,
      "step": 5430
    },
    {
      "epoch": 0.04500144765686396,
      "grad_norm": 2583.996826171875,
      "learning_rate": 4.395959595959596e-06,
      "loss": 243.6479,
      "step": 5440
    },
    {
      "epoch": 0.0450841709062332,
      "grad_norm": 940.0198974609375,
      "learning_rate": 4.4040404040404044e-06,
      "loss": 209.4657,
      "step": 5450
    },
    {
      "epoch": 0.04516689415560243,
      "grad_norm": 897.3545532226562,
      "learning_rate": 4.412121212121213e-06,
      "loss": 244.724,
      "step": 5460
    },
    {
      "epoch": 0.045249617404971666,
      "grad_norm": 1344.112060546875,
      "learning_rate": 4.4202020202020205e-06,
      "loss": 210.6839,
      "step": 5470
    },
    {
      "epoch": 0.0453323406543409,
      "grad_norm": 2638.66357421875,
      "learning_rate": 4.428282828282829e-06,
      "loss": 281.9325,
      "step": 5480
    },
    {
      "epoch": 0.04541506390371014,
      "grad_norm": 1480.95068359375,
      "learning_rate": 4.436363636363637e-06,
      "loss": 227.8617,
      "step": 5490
    },
    {
      "epoch": 0.04549778715307937,
      "grad_norm": 1484.61962890625,
      "learning_rate": 4.444444444444444e-06,
      "loss": 307.6048,
      "step": 5500
    },
    {
      "epoch": 0.04558051040244861,
      "grad_norm": 921.8578491210938,
      "learning_rate": 4.452525252525253e-06,
      "loss": 283.7532,
      "step": 5510
    },
    {
      "epoch": 0.04566323365181784,
      "grad_norm": 1015.0638427734375,
      "learning_rate": 4.460606060606061e-06,
      "loss": 232.6935,
      "step": 5520
    },
    {
      "epoch": 0.04574595690118708,
      "grad_norm": 1154.900146484375,
      "learning_rate": 4.468686868686869e-06,
      "loss": 237.65,
      "step": 5530
    },
    {
      "epoch": 0.045828680150556314,
      "grad_norm": 1067.7034912109375,
      "learning_rate": 4.476767676767677e-06,
      "loss": 178.2383,
      "step": 5540
    },
    {
      "epoch": 0.04591140339992555,
      "grad_norm": 1582.29541015625,
      "learning_rate": 4.4848484848484855e-06,
      "loss": 208.2042,
      "step": 5550
    },
    {
      "epoch": 0.04599412664929478,
      "grad_norm": 1326.168212890625,
      "learning_rate": 4.492929292929293e-06,
      "loss": 206.2426,
      "step": 5560
    },
    {
      "epoch": 0.04607684989866402,
      "grad_norm": 1166.575927734375,
      "learning_rate": 4.501010101010102e-06,
      "loss": 246.9012,
      "step": 5570
    },
    {
      "epoch": 0.046159573148033255,
      "grad_norm": 1051.04345703125,
      "learning_rate": 4.50909090909091e-06,
      "loss": 220.1656,
      "step": 5580
    },
    {
      "epoch": 0.04624229639740249,
      "grad_norm": 994.1414794921875,
      "learning_rate": 4.517171717171718e-06,
      "loss": 246.7432,
      "step": 5590
    },
    {
      "epoch": 0.04632501964677173,
      "grad_norm": 1792.089111328125,
      "learning_rate": 4.525252525252526e-06,
      "loss": 195.5538,
      "step": 5600
    },
    {
      "epoch": 0.04640774289614096,
      "grad_norm": 889.2850341796875,
      "learning_rate": 4.533333333333334e-06,
      "loss": 205.7484,
      "step": 5610
    },
    {
      "epoch": 0.046490466145510195,
      "grad_norm": 986.1232299804688,
      "learning_rate": 4.541414141414141e-06,
      "loss": 259.9392,
      "step": 5620
    },
    {
      "epoch": 0.04657318939487943,
      "grad_norm": 1162.5496826171875,
      "learning_rate": 4.54949494949495e-06,
      "loss": 215.2503,
      "step": 5630
    },
    {
      "epoch": 0.04665591264424867,
      "grad_norm": 1568.9271240234375,
      "learning_rate": 4.557575757575758e-06,
      "loss": 198.5209,
      "step": 5640
    },
    {
      "epoch": 0.0467386358936179,
      "grad_norm": 1163.0565185546875,
      "learning_rate": 4.565656565656566e-06,
      "loss": 199.4977,
      "step": 5650
    },
    {
      "epoch": 0.046821359142987136,
      "grad_norm": 918.8221435546875,
      "learning_rate": 4.573737373737374e-06,
      "loss": 292.0952,
      "step": 5660
    },
    {
      "epoch": 0.04690408239235637,
      "grad_norm": 1420.076171875,
      "learning_rate": 4.581818181818183e-06,
      "loss": 273.1453,
      "step": 5670
    },
    {
      "epoch": 0.04698680564172561,
      "grad_norm": 751.3970947265625,
      "learning_rate": 4.58989898989899e-06,
      "loss": 227.5678,
      "step": 5680
    },
    {
      "epoch": 0.04706952889109484,
      "grad_norm": 4106.84814453125,
      "learning_rate": 4.597979797979799e-06,
      "loss": 257.4646,
      "step": 5690
    },
    {
      "epoch": 0.04715225214046408,
      "grad_norm": 1618.319091796875,
      "learning_rate": 4.606060606060606e-06,
      "loss": 242.2784,
      "step": 5700
    },
    {
      "epoch": 0.04723497538983331,
      "grad_norm": 1682.986083984375,
      "learning_rate": 4.614141414141414e-06,
      "loss": 270.8784,
      "step": 5710
    },
    {
      "epoch": 0.04731769863920255,
      "grad_norm": 2293.774658203125,
      "learning_rate": 4.622222222222222e-06,
      "loss": 310.2911,
      "step": 5720
    },
    {
      "epoch": 0.047400421888571784,
      "grad_norm": 896.1268920898438,
      "learning_rate": 4.630303030303031e-06,
      "loss": 262.9586,
      "step": 5730
    },
    {
      "epoch": 0.04748314513794102,
      "grad_norm": 1522.369384765625,
      "learning_rate": 4.6383838383838384e-06,
      "loss": 200.667,
      "step": 5740
    },
    {
      "epoch": 0.04756586838731025,
      "grad_norm": 1271.126220703125,
      "learning_rate": 4.646464646464647e-06,
      "loss": 211.7462,
      "step": 5750
    },
    {
      "epoch": 0.04764859163667949,
      "grad_norm": 1342.4813232421875,
      "learning_rate": 4.654545454545455e-06,
      "loss": 271.1495,
      "step": 5760
    },
    {
      "epoch": 0.047731314886048724,
      "grad_norm": 1042.1790771484375,
      "learning_rate": 4.662626262626263e-06,
      "loss": 207.4777,
      "step": 5770
    },
    {
      "epoch": 0.04781403813541796,
      "grad_norm": 1017.2898559570312,
      "learning_rate": 4.670707070707071e-06,
      "loss": 272.7402,
      "step": 5780
    },
    {
      "epoch": 0.04789676138478719,
      "grad_norm": 1240.869140625,
      "learning_rate": 4.678787878787879e-06,
      "loss": 224.074,
      "step": 5790
    },
    {
      "epoch": 0.04797948463415643,
      "grad_norm": 1895.0374755859375,
      "learning_rate": 4.6868686868686874e-06,
      "loss": 348.4789,
      "step": 5800
    },
    {
      "epoch": 0.048062207883525665,
      "grad_norm": 1665.503662109375,
      "learning_rate": 4.694949494949496e-06,
      "loss": 287.3858,
      "step": 5810
    },
    {
      "epoch": 0.0481449311328949,
      "grad_norm": 888.2938842773438,
      "learning_rate": 4.7030303030303035e-06,
      "loss": 186.7068,
      "step": 5820
    },
    {
      "epoch": 0.04822765438226414,
      "grad_norm": 1077.885009765625,
      "learning_rate": 4.711111111111111e-06,
      "loss": 188.8208,
      "step": 5830
    },
    {
      "epoch": 0.04831037763163337,
      "grad_norm": 1367.186279296875,
      "learning_rate": 4.7191919191919195e-06,
      "loss": 235.7021,
      "step": 5840
    },
    {
      "epoch": 0.048393100881002606,
      "grad_norm": 880.5682373046875,
      "learning_rate": 4.727272727272728e-06,
      "loss": 150.1969,
      "step": 5850
    },
    {
      "epoch": 0.04847582413037184,
      "grad_norm": 1563.840576171875,
      "learning_rate": 4.735353535353536e-06,
      "loss": 213.3092,
      "step": 5860
    },
    {
      "epoch": 0.04855854737974108,
      "grad_norm": 1928.1456298828125,
      "learning_rate": 4.743434343434344e-06,
      "loss": 258.8421,
      "step": 5870
    },
    {
      "epoch": 0.04864127062911031,
      "grad_norm": 1255.3079833984375,
      "learning_rate": 4.751515151515152e-06,
      "loss": 242.0784,
      "step": 5880
    },
    {
      "epoch": 0.048723993878479546,
      "grad_norm": 1542.299072265625,
      "learning_rate": 4.75959595959596e-06,
      "loss": 303.837,
      "step": 5890
    },
    {
      "epoch": 0.04880671712784878,
      "grad_norm": 2009.2406005859375,
      "learning_rate": 4.7676767676767685e-06,
      "loss": 155.5049,
      "step": 5900
    },
    {
      "epoch": 0.04888944037721802,
      "grad_norm": 1701.25830078125,
      "learning_rate": 4.775757575757576e-06,
      "loss": 289.9675,
      "step": 5910
    },
    {
      "epoch": 0.04897216362658725,
      "grad_norm": 1794.075439453125,
      "learning_rate": 4.783838383838385e-06,
      "loss": 206.6406,
      "step": 5920
    },
    {
      "epoch": 0.04905488687595649,
      "grad_norm": 2225.670654296875,
      "learning_rate": 4.791919191919192e-06,
      "loss": 241.3904,
      "step": 5930
    },
    {
      "epoch": 0.04913761012532572,
      "grad_norm": 1904.87939453125,
      "learning_rate": 4.800000000000001e-06,
      "loss": 245.3052,
      "step": 5940
    },
    {
      "epoch": 0.04922033337469496,
      "grad_norm": 1126.554931640625,
      "learning_rate": 4.808080808080808e-06,
      "loss": 188.594,
      "step": 5950
    },
    {
      "epoch": 0.049303056624064194,
      "grad_norm": 691.1046142578125,
      "learning_rate": 4.816161616161617e-06,
      "loss": 248.3712,
      "step": 5960
    },
    {
      "epoch": 0.04938577987343343,
      "grad_norm": 1169.880859375,
      "learning_rate": 4.824242424242424e-06,
      "loss": 245.4409,
      "step": 5970
    },
    {
      "epoch": 0.04946850312280266,
      "grad_norm": 1042.1029052734375,
      "learning_rate": 4.832323232323233e-06,
      "loss": 202.4587,
      "step": 5980
    },
    {
      "epoch": 0.0495512263721719,
      "grad_norm": 1452.813720703125,
      "learning_rate": 4.840404040404041e-06,
      "loss": 257.6912,
      "step": 5990
    },
    {
      "epoch": 0.049633949621541135,
      "grad_norm": 1247.24609375,
      "learning_rate": 4.848484848484849e-06,
      "loss": 210.8361,
      "step": 6000
    },
    {
      "epoch": 0.04971667287091037,
      "grad_norm": 1635.0625,
      "learning_rate": 4.856565656565657e-06,
      "loss": 241.0676,
      "step": 6010
    },
    {
      "epoch": 0.0497993961202796,
      "grad_norm": 1609.6522216796875,
      "learning_rate": 4.864646464646466e-06,
      "loss": 188.2872,
      "step": 6020
    },
    {
      "epoch": 0.04988211936964884,
      "grad_norm": 1085.433837890625,
      "learning_rate": 4.872727272727273e-06,
      "loss": 194.121,
      "step": 6030
    },
    {
      "epoch": 0.049964842619018075,
      "grad_norm": 1211.4505615234375,
      "learning_rate": 4.880808080808081e-06,
      "loss": 237.9093,
      "step": 6040
    },
    {
      "epoch": 0.05004756586838731,
      "grad_norm": 1017.5370483398438,
      "learning_rate": 4.888888888888889e-06,
      "loss": 298.4579,
      "step": 6050
    },
    {
      "epoch": 0.05013028911775654,
      "grad_norm": 1557.1414794921875,
      "learning_rate": 4.896969696969697e-06,
      "loss": 254.7002,
      "step": 6060
    },
    {
      "epoch": 0.05021301236712578,
      "grad_norm": 966.5237426757812,
      "learning_rate": 4.905050505050505e-06,
      "loss": 218.6844,
      "step": 6070
    },
    {
      "epoch": 0.050295735616495016,
      "grad_norm": 1119.049072265625,
      "learning_rate": 4.913131313131314e-06,
      "loss": 185.6555,
      "step": 6080
    },
    {
      "epoch": 0.05037845886586425,
      "grad_norm": 813.0082397460938,
      "learning_rate": 4.9212121212121214e-06,
      "loss": 193.8167,
      "step": 6090
    },
    {
      "epoch": 0.05046118211523349,
      "grad_norm": 1165.0540771484375,
      "learning_rate": 4.92929292929293e-06,
      "loss": 270.3936,
      "step": 6100
    },
    {
      "epoch": 0.05054390536460272,
      "grad_norm": 774.7178344726562,
      "learning_rate": 4.937373737373738e-06,
      "loss": 187.8722,
      "step": 6110
    },
    {
      "epoch": 0.050626628613971957,
      "grad_norm": 1326.367919921875,
      "learning_rate": 4.945454545454546e-06,
      "loss": 243.3256,
      "step": 6120
    },
    {
      "epoch": 0.05070935186334119,
      "grad_norm": 1486.3304443359375,
      "learning_rate": 4.953535353535354e-06,
      "loss": 186.0064,
      "step": 6130
    },
    {
      "epoch": 0.05079207511271043,
      "grad_norm": 1937.40234375,
      "learning_rate": 4.961616161616162e-06,
      "loss": 246.7979,
      "step": 6140
    },
    {
      "epoch": 0.050874798362079664,
      "grad_norm": 1517.21923828125,
      "learning_rate": 4.9696969696969696e-06,
      "loss": 231.4793,
      "step": 6150
    },
    {
      "epoch": 0.0509575216114489,
      "grad_norm": 5960.52197265625,
      "learning_rate": 4.977777777777778e-06,
      "loss": 197.895,
      "step": 6160
    },
    {
      "epoch": 0.05104024486081813,
      "grad_norm": 1942.2510986328125,
      "learning_rate": 4.9858585858585865e-06,
      "loss": 300.5423,
      "step": 6170
    },
    {
      "epoch": 0.05112296811018737,
      "grad_norm": 1276.1055908203125,
      "learning_rate": 4.993939393939394e-06,
      "loss": 288.2531,
      "step": 6180
    },
    {
      "epoch": 0.051205691359556604,
      "grad_norm": 2154.565673828125,
      "learning_rate": 5.0020202020202025e-06,
      "loss": 259.9872,
      "step": 6190
    },
    {
      "epoch": 0.05128841460892584,
      "grad_norm": 961.3928833007812,
      "learning_rate": 5.010101010101011e-06,
      "loss": 177.8174,
      "step": 6200
    },
    {
      "epoch": 0.05137113785829507,
      "grad_norm": 1438.574462890625,
      "learning_rate": 5.0181818181818186e-06,
      "loss": 197.5066,
      "step": 6210
    },
    {
      "epoch": 0.05145386110766431,
      "grad_norm": 766.4448852539062,
      "learning_rate": 5.026262626262627e-06,
      "loss": 257.3491,
      "step": 6220
    },
    {
      "epoch": 0.051536584357033545,
      "grad_norm": 793.5451049804688,
      "learning_rate": 5.034343434343435e-06,
      "loss": 184.9512,
      "step": 6230
    },
    {
      "epoch": 0.05161930760640278,
      "grad_norm": 869.6885375976562,
      "learning_rate": 5.042424242424243e-06,
      "loss": 253.1606,
      "step": 6240
    },
    {
      "epoch": 0.05170203085577201,
      "grad_norm": 1229.890869140625,
      "learning_rate": 5.0505050505050515e-06,
      "loss": 184.6765,
      "step": 6250
    },
    {
      "epoch": 0.05178475410514125,
      "grad_norm": 804.93994140625,
      "learning_rate": 5.058585858585859e-06,
      "loss": 162.1275,
      "step": 6260
    },
    {
      "epoch": 0.051867477354510486,
      "grad_norm": 1098.615478515625,
      "learning_rate": 5.0666666666666676e-06,
      "loss": 174.5565,
      "step": 6270
    },
    {
      "epoch": 0.05195020060387972,
      "grad_norm": 570.5386352539062,
      "learning_rate": 5.074747474747476e-06,
      "loss": 182.8037,
      "step": 6280
    },
    {
      "epoch": 0.05203292385324895,
      "grad_norm": 3363.004150390625,
      "learning_rate": 5.082828282828284e-06,
      "loss": 265.7634,
      "step": 6290
    },
    {
      "epoch": 0.05211564710261819,
      "grad_norm": 2040.7962646484375,
      "learning_rate": 5.090909090909091e-06,
      "loss": 240.1976,
      "step": 6300
    },
    {
      "epoch": 0.052198370351987426,
      "grad_norm": 1316.55908203125,
      "learning_rate": 5.098989898989899e-06,
      "loss": 249.6734,
      "step": 6310
    },
    {
      "epoch": 0.05228109360135666,
      "grad_norm": 2261.989013671875,
      "learning_rate": 5.107070707070707e-06,
      "loss": 208.5467,
      "step": 6320
    },
    {
      "epoch": 0.0523638168507259,
      "grad_norm": 1246.7120361328125,
      "learning_rate": 5.115151515151515e-06,
      "loss": 207.1795,
      "step": 6330
    },
    {
      "epoch": 0.05244654010009513,
      "grad_norm": 1235.8924560546875,
      "learning_rate": 5.123232323232323e-06,
      "loss": 231.5552,
      "step": 6340
    },
    {
      "epoch": 0.05252926334946437,
      "grad_norm": 805.1367797851562,
      "learning_rate": 5.131313131313132e-06,
      "loss": 217.3461,
      "step": 6350
    },
    {
      "epoch": 0.0526119865988336,
      "grad_norm": 902.8432006835938,
      "learning_rate": 5.139393939393939e-06,
      "loss": 250.8667,
      "step": 6360
    },
    {
      "epoch": 0.05269470984820284,
      "grad_norm": 1314.75634765625,
      "learning_rate": 5.147474747474748e-06,
      "loss": 188.1431,
      "step": 6370
    },
    {
      "epoch": 0.052777433097572074,
      "grad_norm": 1282.649169921875,
      "learning_rate": 5.155555555555556e-06,
      "loss": 198.1481,
      "step": 6380
    },
    {
      "epoch": 0.05286015634694131,
      "grad_norm": 1167.904052734375,
      "learning_rate": 5.163636363636364e-06,
      "loss": 192.307,
      "step": 6390
    },
    {
      "epoch": 0.05294287959631054,
      "grad_norm": 686.8750610351562,
      "learning_rate": 5.171717171717172e-06,
      "loss": 225.3385,
      "step": 6400
    },
    {
      "epoch": 0.05302560284567978,
      "grad_norm": 1144.7125244140625,
      "learning_rate": 5.17979797979798e-06,
      "loss": 188.2769,
      "step": 6410
    },
    {
      "epoch": 0.053108326095049015,
      "grad_norm": 1010.5237426757812,
      "learning_rate": 5.187878787878788e-06,
      "loss": 225.7106,
      "step": 6420
    },
    {
      "epoch": 0.05319104934441825,
      "grad_norm": 1336.3463134765625,
      "learning_rate": 5.195959595959597e-06,
      "loss": 208.7707,
      "step": 6430
    },
    {
      "epoch": 0.05327377259378748,
      "grad_norm": 1576.3480224609375,
      "learning_rate": 5.204040404040404e-06,
      "loss": 236.1072,
      "step": 6440
    },
    {
      "epoch": 0.05335649584315672,
      "grad_norm": 1445.076416015625,
      "learning_rate": 5.212121212121213e-06,
      "loss": 271.4749,
      "step": 6450
    },
    {
      "epoch": 0.053439219092525955,
      "grad_norm": 895.9810791015625,
      "learning_rate": 5.220202020202021e-06,
      "loss": 258.855,
      "step": 6460
    },
    {
      "epoch": 0.05352194234189519,
      "grad_norm": 952.8675537109375,
      "learning_rate": 5.228282828282829e-06,
      "loss": 175.3387,
      "step": 6470
    },
    {
      "epoch": 0.05360466559126442,
      "grad_norm": 1066.5716552734375,
      "learning_rate": 5.236363636363637e-06,
      "loss": 218.2292,
      "step": 6480
    },
    {
      "epoch": 0.05368738884063366,
      "grad_norm": 849.3695678710938,
      "learning_rate": 5.244444444444445e-06,
      "loss": 221.0137,
      "step": 6490
    },
    {
      "epoch": 0.053770112090002896,
      "grad_norm": 1675.2149658203125,
      "learning_rate": 5.252525252525253e-06,
      "loss": 223.2969,
      "step": 6500
    },
    {
      "epoch": 0.05385283533937213,
      "grad_norm": 1693.541748046875,
      "learning_rate": 5.26060606060606e-06,
      "loss": 229.6179,
      "step": 6510
    },
    {
      "epoch": 0.05393555858874136,
      "grad_norm": 1324.1651611328125,
      "learning_rate": 5.268686868686869e-06,
      "loss": 244.8779,
      "step": 6520
    },
    {
      "epoch": 0.0540182818381106,
      "grad_norm": 1403.1368408203125,
      "learning_rate": 5.276767676767677e-06,
      "loss": 268.8457,
      "step": 6530
    },
    {
      "epoch": 0.054101005087479836,
      "grad_norm": 1165.638671875,
      "learning_rate": 5.284848484848485e-06,
      "loss": 239.8147,
      "step": 6540
    },
    {
      "epoch": 0.05418372833684907,
      "grad_norm": 995.2945556640625,
      "learning_rate": 5.292929292929293e-06,
      "loss": 229.9961,
      "step": 6550
    },
    {
      "epoch": 0.0542664515862183,
      "grad_norm": 1505.5985107421875,
      "learning_rate": 5.3010101010101016e-06,
      "loss": 223.1589,
      "step": 6560
    },
    {
      "epoch": 0.054349174835587544,
      "grad_norm": 1667.1300048828125,
      "learning_rate": 5.309090909090909e-06,
      "loss": 217.6828,
      "step": 6570
    },
    {
      "epoch": 0.05443189808495678,
      "grad_norm": 1639.0364990234375,
      "learning_rate": 5.317171717171718e-06,
      "loss": 205.0005,
      "step": 6580
    },
    {
      "epoch": 0.05451462133432601,
      "grad_norm": 1414.21240234375,
      "learning_rate": 5.325252525252525e-06,
      "loss": 215.9356,
      "step": 6590
    },
    {
      "epoch": 0.05459734458369525,
      "grad_norm": 586.2081298828125,
      "learning_rate": 5.333333333333334e-06,
      "loss": 227.373,
      "step": 6600
    },
    {
      "epoch": 0.054680067833064484,
      "grad_norm": 2097.269775390625,
      "learning_rate": 5.341414141414142e-06,
      "loss": 234.4245,
      "step": 6610
    },
    {
      "epoch": 0.05476279108243372,
      "grad_norm": 1416.9189453125,
      "learning_rate": 5.34949494949495e-06,
      "loss": 262.4427,
      "step": 6620
    },
    {
      "epoch": 0.05484551433180295,
      "grad_norm": 1144.749755859375,
      "learning_rate": 5.357575757575758e-06,
      "loss": 201.3394,
      "step": 6630
    },
    {
      "epoch": 0.05492823758117219,
      "grad_norm": 1287.0030517578125,
      "learning_rate": 5.365656565656567e-06,
      "loss": 281.6339,
      "step": 6640
    },
    {
      "epoch": 0.055010960830541425,
      "grad_norm": 1555.04150390625,
      "learning_rate": 5.373737373737374e-06,
      "loss": 225.9727,
      "step": 6650
    },
    {
      "epoch": 0.05509368407991066,
      "grad_norm": 1129.1629638671875,
      "learning_rate": 5.381818181818183e-06,
      "loss": 159.6996,
      "step": 6660
    },
    {
      "epoch": 0.05517640732927989,
      "grad_norm": 1159.3424072265625,
      "learning_rate": 5.38989898989899e-06,
      "loss": 226.1631,
      "step": 6670
    },
    {
      "epoch": 0.05525913057864913,
      "grad_norm": 1065.8494873046875,
      "learning_rate": 5.397979797979799e-06,
      "loss": 213.2366,
      "step": 6680
    },
    {
      "epoch": 0.055341853828018366,
      "grad_norm": 801.6089477539062,
      "learning_rate": 5.406060606060607e-06,
      "loss": 180.9512,
      "step": 6690
    },
    {
      "epoch": 0.0554245770773876,
      "grad_norm": 1254.9515380859375,
      "learning_rate": 5.414141414141415e-06,
      "loss": 223.2975,
      "step": 6700
    },
    {
      "epoch": 0.05550730032675683,
      "grad_norm": 1734.849609375,
      "learning_rate": 5.422222222222223e-06,
      "loss": 179.4333,
      "step": 6710
    },
    {
      "epoch": 0.05559002357612607,
      "grad_norm": 1987.9564208984375,
      "learning_rate": 5.430303030303032e-06,
      "loss": 248.2462,
      "step": 6720
    },
    {
      "epoch": 0.055672746825495306,
      "grad_norm": 1181.2037353515625,
      "learning_rate": 5.438383838383838e-06,
      "loss": 182.878,
      "step": 6730
    },
    {
      "epoch": 0.05575547007486454,
      "grad_norm": 1901.66015625,
      "learning_rate": 5.446464646464647e-06,
      "loss": 251.8006,
      "step": 6740
    },
    {
      "epoch": 0.05583819332423377,
      "grad_norm": 632.8333740234375,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 196.6042,
      "step": 6750
    },
    {
      "epoch": 0.05592091657360301,
      "grad_norm": 604.5985717773438,
      "learning_rate": 5.462626262626263e-06,
      "loss": 177.947,
      "step": 6760
    },
    {
      "epoch": 0.05600363982297225,
      "grad_norm": 1062.9954833984375,
      "learning_rate": 5.4707070707070705e-06,
      "loss": 312.2176,
      "step": 6770
    },
    {
      "epoch": 0.05608636307234148,
      "grad_norm": 1964.1070556640625,
      "learning_rate": 5.478787878787879e-06,
      "loss": 192.9068,
      "step": 6780
    },
    {
      "epoch": 0.056169086321710714,
      "grad_norm": 1671.50146484375,
      "learning_rate": 5.486868686868687e-06,
      "loss": 236.6793,
      "step": 6790
    },
    {
      "epoch": 0.056251809571079954,
      "grad_norm": 1363.7489013671875,
      "learning_rate": 5.494949494949495e-06,
      "loss": 207.5258,
      "step": 6800
    },
    {
      "epoch": 0.05633453282044919,
      "grad_norm": 1217.836669921875,
      "learning_rate": 5.5030303030303034e-06,
      "loss": 185.7013,
      "step": 6810
    },
    {
      "epoch": 0.05641725606981842,
      "grad_norm": 1585.8988037109375,
      "learning_rate": 5.511111111111112e-06,
      "loss": 217.5803,
      "step": 6820
    },
    {
      "epoch": 0.056499979319187654,
      "grad_norm": 983.3916625976562,
      "learning_rate": 5.5191919191919195e-06,
      "loss": 224.2803,
      "step": 6830
    },
    {
      "epoch": 0.056582702568556895,
      "grad_norm": 1328.7664794921875,
      "learning_rate": 5.527272727272728e-06,
      "loss": 213.4681,
      "step": 6840
    },
    {
      "epoch": 0.05666542581792613,
      "grad_norm": 908.5962524414062,
      "learning_rate": 5.5353535353535355e-06,
      "loss": 191.1099,
      "step": 6850
    },
    {
      "epoch": 0.05674814906729536,
      "grad_norm": 965.225341796875,
      "learning_rate": 5.543434343434344e-06,
      "loss": 180.1947,
      "step": 6860
    },
    {
      "epoch": 0.0568308723166646,
      "grad_norm": 1269.7061767578125,
      "learning_rate": 5.5515151515151524e-06,
      "loss": 235.1135,
      "step": 6870
    },
    {
      "epoch": 0.056913595566033835,
      "grad_norm": 1158.512939453125,
      "learning_rate": 5.55959595959596e-06,
      "loss": 215.2743,
      "step": 6880
    },
    {
      "epoch": 0.05699631881540307,
      "grad_norm": 1231.1844482421875,
      "learning_rate": 5.5676767676767685e-06,
      "loss": 327.6613,
      "step": 6890
    },
    {
      "epoch": 0.0570790420647723,
      "grad_norm": 1180.1192626953125,
      "learning_rate": 5.575757575757577e-06,
      "loss": 228.0458,
      "step": 6900
    },
    {
      "epoch": 0.05716176531414154,
      "grad_norm": 952.682861328125,
      "learning_rate": 5.5838383838383845e-06,
      "loss": 250.6257,
      "step": 6910
    },
    {
      "epoch": 0.057244488563510776,
      "grad_norm": 16013.0146484375,
      "learning_rate": 5.591919191919193e-06,
      "loss": 250.4285,
      "step": 6920
    },
    {
      "epoch": 0.05732721181288001,
      "grad_norm": 1266.31787109375,
      "learning_rate": 5.600000000000001e-06,
      "loss": 213.8897,
      "step": 6930
    },
    {
      "epoch": 0.05740993506224924,
      "grad_norm": 1832.4088134765625,
      "learning_rate": 5.608080808080808e-06,
      "loss": 218.5001,
      "step": 6940
    },
    {
      "epoch": 0.05749265831161848,
      "grad_norm": 1213.3984375,
      "learning_rate": 5.616161616161616e-06,
      "loss": 211.7139,
      "step": 6950
    },
    {
      "epoch": 0.057575381560987716,
      "grad_norm": 1369.9437255859375,
      "learning_rate": 5.624242424242424e-06,
      "loss": 192.9413,
      "step": 6960
    },
    {
      "epoch": 0.05765810481035695,
      "grad_norm": 1018.5819091796875,
      "learning_rate": 5.632323232323233e-06,
      "loss": 150.2073,
      "step": 6970
    },
    {
      "epoch": 0.05774082805972618,
      "grad_norm": 1435.1353759765625,
      "learning_rate": 5.64040404040404e-06,
      "loss": 187.5981,
      "step": 6980
    },
    {
      "epoch": 0.057823551309095424,
      "grad_norm": 1339.41650390625,
      "learning_rate": 5.648484848484849e-06,
      "loss": 165.7803,
      "step": 6990
    },
    {
      "epoch": 0.05790627455846466,
      "grad_norm": 3150.48974609375,
      "learning_rate": 5.656565656565657e-06,
      "loss": 183.762,
      "step": 7000
    },
    {
      "epoch": 0.05798899780783389,
      "grad_norm": 1491.4512939453125,
      "learning_rate": 5.664646464646465e-06,
      "loss": 205.812,
      "step": 7010
    },
    {
      "epoch": 0.058071721057203124,
      "grad_norm": 1045.4873046875,
      "learning_rate": 5.672727272727273e-06,
      "loss": 222.1431,
      "step": 7020
    },
    {
      "epoch": 0.058154444306572364,
      "grad_norm": 1604.70556640625,
      "learning_rate": 5.680808080808081e-06,
      "loss": 312.9387,
      "step": 7030
    },
    {
      "epoch": 0.0582371675559416,
      "grad_norm": 794.0164184570312,
      "learning_rate": 5.688888888888889e-06,
      "loss": 148.5473,
      "step": 7040
    },
    {
      "epoch": 0.05831989080531083,
      "grad_norm": 1710.36328125,
      "learning_rate": 5.696969696969698e-06,
      "loss": 187.5921,
      "step": 7050
    },
    {
      "epoch": 0.058402614054680065,
      "grad_norm": 1974.2613525390625,
      "learning_rate": 5.705050505050505e-06,
      "loss": 192.2352,
      "step": 7060
    },
    {
      "epoch": 0.058485337304049305,
      "grad_norm": 1533.34619140625,
      "learning_rate": 5.713131313131314e-06,
      "loss": 189.592,
      "step": 7070
    },
    {
      "epoch": 0.05856806055341854,
      "grad_norm": 1127.27392578125,
      "learning_rate": 5.721212121212122e-06,
      "loss": 211.6275,
      "step": 7080
    },
    {
      "epoch": 0.05865078380278777,
      "grad_norm": 1902.27001953125,
      "learning_rate": 5.72929292929293e-06,
      "loss": 189.9227,
      "step": 7090
    },
    {
      "epoch": 0.05873350705215701,
      "grad_norm": 1564.533203125,
      "learning_rate": 5.737373737373738e-06,
      "loss": 231.5597,
      "step": 7100
    },
    {
      "epoch": 0.058816230301526246,
      "grad_norm": 2943.590576171875,
      "learning_rate": 5.745454545454546e-06,
      "loss": 177.4852,
      "step": 7110
    },
    {
      "epoch": 0.05889895355089548,
      "grad_norm": 1309.1622314453125,
      "learning_rate": 5.753535353535354e-06,
      "loss": 165.6322,
      "step": 7120
    },
    {
      "epoch": 0.05898167680026471,
      "grad_norm": 1501.813720703125,
      "learning_rate": 5.761616161616163e-06,
      "loss": 189.5367,
      "step": 7130
    },
    {
      "epoch": 0.05906440004963395,
      "grad_norm": 1574.177978515625,
      "learning_rate": 5.76969696969697e-06,
      "loss": 234.3292,
      "step": 7140
    },
    {
      "epoch": 0.059147123299003186,
      "grad_norm": 1052.1170654296875,
      "learning_rate": 5.777777777777778e-06,
      "loss": 214.4601,
      "step": 7150
    },
    {
      "epoch": 0.05922984654837242,
      "grad_norm": 1018.6898803710938,
      "learning_rate": 5.785858585858586e-06,
      "loss": 156.9678,
      "step": 7160
    },
    {
      "epoch": 0.05931256979774165,
      "grad_norm": 1917.61865234375,
      "learning_rate": 5.793939393939394e-06,
      "loss": 233.6657,
      "step": 7170
    },
    {
      "epoch": 0.05939529304711089,
      "grad_norm": 1654.3441162109375,
      "learning_rate": 5.8020202020202025e-06,
      "loss": 173.1022,
      "step": 7180
    },
    {
      "epoch": 0.05947801629648013,
      "grad_norm": 1147.4951171875,
      "learning_rate": 5.81010101010101e-06,
      "loss": 209.7125,
      "step": 7190
    },
    {
      "epoch": 0.05956073954584936,
      "grad_norm": 1505.357177734375,
      "learning_rate": 5.8181818181818185e-06,
      "loss": 239.9892,
      "step": 7200
    },
    {
      "epoch": 0.059643462795218594,
      "grad_norm": 3524.7958984375,
      "learning_rate": 5.826262626262626e-06,
      "loss": 199.2161,
      "step": 7210
    },
    {
      "epoch": 0.059726186044587834,
      "grad_norm": 2917.4033203125,
      "learning_rate": 5.834343434343435e-06,
      "loss": 282.2665,
      "step": 7220
    },
    {
      "epoch": 0.05980890929395707,
      "grad_norm": 976.106689453125,
      "learning_rate": 5.842424242424243e-06,
      "loss": 220.0685,
      "step": 7230
    },
    {
      "epoch": 0.0598916325433263,
      "grad_norm": 1532.435791015625,
      "learning_rate": 5.850505050505051e-06,
      "loss": 247.5289,
      "step": 7240
    },
    {
      "epoch": 0.059974355792695534,
      "grad_norm": 868.5018920898438,
      "learning_rate": 5.858585858585859e-06,
      "loss": 204.2799,
      "step": 7250
    },
    {
      "epoch": 0.060057079042064775,
      "grad_norm": 1144.6495361328125,
      "learning_rate": 5.8666666666666675e-06,
      "loss": 233.1151,
      "step": 7260
    },
    {
      "epoch": 0.06013980229143401,
      "grad_norm": 1262.310302734375,
      "learning_rate": 5.874747474747475e-06,
      "loss": 215.0127,
      "step": 7270
    },
    {
      "epoch": 0.06022252554080324,
      "grad_norm": 1465.82568359375,
      "learning_rate": 5.882828282828284e-06,
      "loss": 214.6379,
      "step": 7280
    },
    {
      "epoch": 0.060305248790172475,
      "grad_norm": 890.1478271484375,
      "learning_rate": 5.890909090909091e-06,
      "loss": 212.3409,
      "step": 7290
    },
    {
      "epoch": 0.060387972039541715,
      "grad_norm": 804.8695068359375,
      "learning_rate": 5.8989898989899e-06,
      "loss": 161.0226,
      "step": 7300
    },
    {
      "epoch": 0.06047069528891095,
      "grad_norm": 1049.8092041015625,
      "learning_rate": 5.907070707070708e-06,
      "loss": 228.0541,
      "step": 7310
    },
    {
      "epoch": 0.06055341853828018,
      "grad_norm": 1314.708984375,
      "learning_rate": 5.915151515151516e-06,
      "loss": 169.0287,
      "step": 7320
    },
    {
      "epoch": 0.060636141787649415,
      "grad_norm": 877.56982421875,
      "learning_rate": 5.923232323232324e-06,
      "loss": 182.722,
      "step": 7330
    },
    {
      "epoch": 0.060718865037018656,
      "grad_norm": 1187.692138671875,
      "learning_rate": 5.9313131313131326e-06,
      "loss": 216.9126,
      "step": 7340
    },
    {
      "epoch": 0.06080158828638789,
      "grad_norm": 910.7957763671875,
      "learning_rate": 5.93939393939394e-06,
      "loss": 156.4217,
      "step": 7350
    },
    {
      "epoch": 0.06088431153575712,
      "grad_norm": 937.8931274414062,
      "learning_rate": 5.947474747474749e-06,
      "loss": 206.2811,
      "step": 7360
    },
    {
      "epoch": 0.06096703478512636,
      "grad_norm": 1068.356201171875,
      "learning_rate": 5.955555555555555e-06,
      "loss": 209.255,
      "step": 7370
    },
    {
      "epoch": 0.061049758034495596,
      "grad_norm": 1415.2979736328125,
      "learning_rate": 5.963636363636364e-06,
      "loss": 180.4899,
      "step": 7380
    },
    {
      "epoch": 0.06113248128386483,
      "grad_norm": 1216.490966796875,
      "learning_rate": 5.9717171717171714e-06,
      "loss": 181.2195,
      "step": 7390
    },
    {
      "epoch": 0.06121520453323406,
      "grad_norm": 1969.770751953125,
      "learning_rate": 5.97979797979798e-06,
      "loss": 185.2952,
      "step": 7400
    },
    {
      "epoch": 0.061297927782603304,
      "grad_norm": 1073.1572265625,
      "learning_rate": 5.987878787878788e-06,
      "loss": 194.6978,
      "step": 7410
    },
    {
      "epoch": 0.06138065103197254,
      "grad_norm": 1329.2269287109375,
      "learning_rate": 5.995959595959596e-06,
      "loss": 225.0991,
      "step": 7420
    },
    {
      "epoch": 0.06146337428134177,
      "grad_norm": 1447.0704345703125,
      "learning_rate": 6.004040404040404e-06,
      "loss": 222.2881,
      "step": 7430
    },
    {
      "epoch": 0.061546097530711004,
      "grad_norm": 1847.17578125,
      "learning_rate": 6.012121212121213e-06,
      "loss": 250.8386,
      "step": 7440
    },
    {
      "epoch": 0.061628820780080244,
      "grad_norm": 1011.4767456054688,
      "learning_rate": 6.0202020202020204e-06,
      "loss": 217.8489,
      "step": 7450
    },
    {
      "epoch": 0.06171154402944948,
      "grad_norm": 905.0313110351562,
      "learning_rate": 6.028282828282829e-06,
      "loss": 176.8829,
      "step": 7460
    },
    {
      "epoch": 0.06179426727881871,
      "grad_norm": 1627.3341064453125,
      "learning_rate": 6.0363636363636365e-06,
      "loss": 245.858,
      "step": 7470
    },
    {
      "epoch": 0.061876990528187945,
      "grad_norm": 1333.339111328125,
      "learning_rate": 6.044444444444445e-06,
      "loss": 172.512,
      "step": 7480
    },
    {
      "epoch": 0.061959713777557185,
      "grad_norm": 981.724609375,
      "learning_rate": 6.052525252525253e-06,
      "loss": 188.6218,
      "step": 7490
    },
    {
      "epoch": 0.06204243702692642,
      "grad_norm": 963.2952880859375,
      "learning_rate": 6.060606060606061e-06,
      "loss": 200.9197,
      "step": 7500
    },
    {
      "epoch": 0.06212516027629565,
      "grad_norm": 1976.6124267578125,
      "learning_rate": 6.068686868686869e-06,
      "loss": 206.6058,
      "step": 7510
    },
    {
      "epoch": 0.062207883525664885,
      "grad_norm": 964.7189331054688,
      "learning_rate": 6.076767676767678e-06,
      "loss": 244.5135,
      "step": 7520
    },
    {
      "epoch": 0.062290606775034126,
      "grad_norm": 940.2135009765625,
      "learning_rate": 6.0848484848484855e-06,
      "loss": 286.4893,
      "step": 7530
    },
    {
      "epoch": 0.06237333002440336,
      "grad_norm": 1086.095947265625,
      "learning_rate": 6.092929292929294e-06,
      "loss": 166.7001,
      "step": 7540
    },
    {
      "epoch": 0.06245605327377259,
      "grad_norm": 824.2728881835938,
      "learning_rate": 6.1010101010101015e-06,
      "loss": 178.9928,
      "step": 7550
    },
    {
      "epoch": 0.06253877652314183,
      "grad_norm": 1227.130859375,
      "learning_rate": 6.10909090909091e-06,
      "loss": 188.424,
      "step": 7560
    },
    {
      "epoch": 0.06262149977251107,
      "grad_norm": 943.979736328125,
      "learning_rate": 6.117171717171718e-06,
      "loss": 170.1881,
      "step": 7570
    },
    {
      "epoch": 0.06270422302188029,
      "grad_norm": 893.2919921875,
      "learning_rate": 6.125252525252525e-06,
      "loss": 203.6881,
      "step": 7580
    },
    {
      "epoch": 0.06278694627124953,
      "grad_norm": 1502.2542724609375,
      "learning_rate": 6.133333333333334e-06,
      "loss": 227.4439,
      "step": 7590
    },
    {
      "epoch": 0.06286966952061877,
      "grad_norm": 2379.823974609375,
      "learning_rate": 6.141414141414141e-06,
      "loss": 216.3211,
      "step": 7600
    },
    {
      "epoch": 0.062952392769988,
      "grad_norm": 2946.320068359375,
      "learning_rate": 6.14949494949495e-06,
      "loss": 181.0182,
      "step": 7610
    },
    {
      "epoch": 0.06303511601935724,
      "grad_norm": 1127.286865234375,
      "learning_rate": 6.157575757575758e-06,
      "loss": 219.9298,
      "step": 7620
    },
    {
      "epoch": 0.06311783926872648,
      "grad_norm": 2195.520751953125,
      "learning_rate": 6.165656565656566e-06,
      "loss": 225.9704,
      "step": 7630
    },
    {
      "epoch": 0.06320056251809571,
      "grad_norm": 1115.85107421875,
      "learning_rate": 6.173737373737374e-06,
      "loss": 192.5431,
      "step": 7640
    },
    {
      "epoch": 0.06328328576746495,
      "grad_norm": 1610.82861328125,
      "learning_rate": 6.181818181818182e-06,
      "loss": 234.5798,
      "step": 7650
    },
    {
      "epoch": 0.06336600901683419,
      "grad_norm": 2228.57080078125,
      "learning_rate": 6.18989898989899e-06,
      "loss": 203.8422,
      "step": 7660
    },
    {
      "epoch": 0.06344873226620341,
      "grad_norm": 1072.0853271484375,
      "learning_rate": 6.197979797979799e-06,
      "loss": 136.4943,
      "step": 7670
    },
    {
      "epoch": 0.06353145551557265,
      "grad_norm": 987.9502563476562,
      "learning_rate": 6.206060606060606e-06,
      "loss": 176.214,
      "step": 7680
    },
    {
      "epoch": 0.06361417876494188,
      "grad_norm": 1846.4237060546875,
      "learning_rate": 6.214141414141415e-06,
      "loss": 328.0838,
      "step": 7690
    },
    {
      "epoch": 0.06369690201431112,
      "grad_norm": 1142.6070556640625,
      "learning_rate": 6.222222222222223e-06,
      "loss": 197.7707,
      "step": 7700
    },
    {
      "epoch": 0.06377962526368036,
      "grad_norm": 2267.01904296875,
      "learning_rate": 6.230303030303031e-06,
      "loss": 246.8396,
      "step": 7710
    },
    {
      "epoch": 0.06386234851304959,
      "grad_norm": 1803.31591796875,
      "learning_rate": 6.238383838383839e-06,
      "loss": 156.6524,
      "step": 7720
    },
    {
      "epoch": 0.06394507176241883,
      "grad_norm": 1524.9769287109375,
      "learning_rate": 6.246464646464647e-06,
      "loss": 252.2839,
      "step": 7730
    },
    {
      "epoch": 0.06402779501178807,
      "grad_norm": 1911.2991943359375,
      "learning_rate": 6.254545454545455e-06,
      "loss": 185.8043,
      "step": 7740
    },
    {
      "epoch": 0.0641105182611573,
      "grad_norm": 1456.3634033203125,
      "learning_rate": 6.262626262626264e-06,
      "loss": 189.6922,
      "step": 7750
    },
    {
      "epoch": 0.06419324151052654,
      "grad_norm": 872.6129150390625,
      "learning_rate": 6.270707070707071e-06,
      "loss": 211.5073,
      "step": 7760
    },
    {
      "epoch": 0.06427596475989576,
      "grad_norm": 2161.432861328125,
      "learning_rate": 6.27878787878788e-06,
      "loss": 227.991,
      "step": 7770
    },
    {
      "epoch": 0.064358688009265,
      "grad_norm": 1427.72900390625,
      "learning_rate": 6.286868686868688e-06,
      "loss": 187.7627,
      "step": 7780
    },
    {
      "epoch": 0.06444141125863424,
      "grad_norm": 1048.0509033203125,
      "learning_rate": 6.294949494949495e-06,
      "loss": 216.9241,
      "step": 7790
    },
    {
      "epoch": 0.06452413450800347,
      "grad_norm": 831.1154174804688,
      "learning_rate": 6.303030303030303e-06,
      "loss": 156.4739,
      "step": 7800
    },
    {
      "epoch": 0.06460685775737271,
      "grad_norm": 1421.5413818359375,
      "learning_rate": 6.311111111111111e-06,
      "loss": 240.7841,
      "step": 7810
    },
    {
      "epoch": 0.06468958100674195,
      "grad_norm": 1609.0704345703125,
      "learning_rate": 6.3191919191919195e-06,
      "loss": 202.0292,
      "step": 7820
    },
    {
      "epoch": 0.06477230425611118,
      "grad_norm": 1056.16064453125,
      "learning_rate": 6.327272727272727e-06,
      "loss": 150.6357,
      "step": 7830
    },
    {
      "epoch": 0.06485502750548042,
      "grad_norm": 3895.295166015625,
      "learning_rate": 6.3353535353535355e-06,
      "loss": 166.0538,
      "step": 7840
    },
    {
      "epoch": 0.06493775075484966,
      "grad_norm": 1299.444580078125,
      "learning_rate": 6.343434343434344e-06,
      "loss": 160.5584,
      "step": 7850
    },
    {
      "epoch": 0.06502047400421888,
      "grad_norm": 712.9761352539062,
      "learning_rate": 6.3515151515151516e-06,
      "loss": 197.3979,
      "step": 7860
    },
    {
      "epoch": 0.06510319725358812,
      "grad_norm": 1037.716064453125,
      "learning_rate": 6.35959595959596e-06,
      "loss": 200.6918,
      "step": 7870
    },
    {
      "epoch": 0.06518592050295735,
      "grad_norm": 1494.057373046875,
      "learning_rate": 6.3676767676767685e-06,
      "loss": 238.1615,
      "step": 7880
    },
    {
      "epoch": 0.06526864375232659,
      "grad_norm": 1253.9837646484375,
      "learning_rate": 6.375757575757576e-06,
      "loss": 190.7486,
      "step": 7890
    },
    {
      "epoch": 0.06535136700169583,
      "grad_norm": 1362.12353515625,
      "learning_rate": 6.3838383838383845e-06,
      "loss": 214.7477,
      "step": 7900
    },
    {
      "epoch": 0.06543409025106506,
      "grad_norm": 890.4193725585938,
      "learning_rate": 6.391919191919192e-06,
      "loss": 170.1652,
      "step": 7910
    },
    {
      "epoch": 0.0655168135004343,
      "grad_norm": 1644.9490966796875,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 175.0597,
      "step": 7920
    },
    {
      "epoch": 0.06559953674980354,
      "grad_norm": 777.9044799804688,
      "learning_rate": 6.408080808080809e-06,
      "loss": 238.5318,
      "step": 7930
    },
    {
      "epoch": 0.06568225999917277,
      "grad_norm": 1498.41455078125,
      "learning_rate": 6.416161616161617e-06,
      "loss": 183.5011,
      "step": 7940
    },
    {
      "epoch": 0.065764983248542,
      "grad_norm": 1392.91748046875,
      "learning_rate": 6.424242424242425e-06,
      "loss": 164.5937,
      "step": 7950
    },
    {
      "epoch": 0.06584770649791123,
      "grad_norm": 1324.7021484375,
      "learning_rate": 6.4323232323232335e-06,
      "loss": 230.2011,
      "step": 7960
    },
    {
      "epoch": 0.06593042974728047,
      "grad_norm": 1371.1768798828125,
      "learning_rate": 6.440404040404041e-06,
      "loss": 226.2801,
      "step": 7970
    },
    {
      "epoch": 0.06601315299664971,
      "grad_norm": 2229.455078125,
      "learning_rate": 6.4484848484848496e-06,
      "loss": 196.2276,
      "step": 7980
    },
    {
      "epoch": 0.06609587624601894,
      "grad_norm": 1298.6649169921875,
      "learning_rate": 6.456565656565658e-06,
      "loss": 211.2707,
      "step": 7990
    },
    {
      "epoch": 0.06617859949538818,
      "grad_norm": 784.8013305664062,
      "learning_rate": 6.464646464646466e-06,
      "loss": 207.776,
      "step": 8000
    },
    {
      "epoch": 0.06626132274475742,
      "grad_norm": 911.982666015625,
      "learning_rate": 6.472727272727272e-06,
      "loss": 162.3888,
      "step": 8010
    },
    {
      "epoch": 0.06634404599412665,
      "grad_norm": 1066.7635498046875,
      "learning_rate": 6.480808080808081e-06,
      "loss": 182.5109,
      "step": 8020
    },
    {
      "epoch": 0.06642676924349589,
      "grad_norm": 733.5493774414062,
      "learning_rate": 6.488888888888889e-06,
      "loss": 140.7046,
      "step": 8030
    },
    {
      "epoch": 0.06650949249286511,
      "grad_norm": 1074.85888671875,
      "learning_rate": 6.496969696969697e-06,
      "loss": 182.7351,
      "step": 8040
    },
    {
      "epoch": 0.06659221574223435,
      "grad_norm": 1245.099365234375,
      "learning_rate": 6.505050505050505e-06,
      "loss": 228.4951,
      "step": 8050
    },
    {
      "epoch": 0.0666749389916036,
      "grad_norm": 820.326171875,
      "learning_rate": 6.513131313131314e-06,
      "loss": 172.91,
      "step": 8060
    },
    {
      "epoch": 0.06675766224097282,
      "grad_norm": 1347.561279296875,
      "learning_rate": 6.521212121212121e-06,
      "loss": 175.6294,
      "step": 8070
    },
    {
      "epoch": 0.06684038549034206,
      "grad_norm": 2743.030517578125,
      "learning_rate": 6.52929292929293e-06,
      "loss": 240.5082,
      "step": 8080
    },
    {
      "epoch": 0.0669231087397113,
      "grad_norm": 1014.7551879882812,
      "learning_rate": 6.537373737373737e-06,
      "loss": 181.1287,
      "step": 8090
    },
    {
      "epoch": 0.06700583198908053,
      "grad_norm": 1014.5245971679688,
      "learning_rate": 6.545454545454546e-06,
      "loss": 189.7482,
      "step": 8100
    },
    {
      "epoch": 0.06708855523844977,
      "grad_norm": 778.7886962890625,
      "learning_rate": 6.553535353535354e-06,
      "loss": 170.7764,
      "step": 8110
    },
    {
      "epoch": 0.06717127848781901,
      "grad_norm": 2081.901611328125,
      "learning_rate": 6.561616161616162e-06,
      "loss": 245.8872,
      "step": 8120
    },
    {
      "epoch": 0.06725400173718823,
      "grad_norm": 1881.017333984375,
      "learning_rate": 6.56969696969697e-06,
      "loss": 252.341,
      "step": 8130
    },
    {
      "epoch": 0.06733672498655748,
      "grad_norm": 910.3214111328125,
      "learning_rate": 6.577777777777779e-06,
      "loss": 179.5906,
      "step": 8140
    },
    {
      "epoch": 0.0674194482359267,
      "grad_norm": 1163.5965576171875,
      "learning_rate": 6.585858585858586e-06,
      "loss": 233.8698,
      "step": 8150
    },
    {
      "epoch": 0.06750217148529594,
      "grad_norm": 884.6065673828125,
      "learning_rate": 6.593939393939395e-06,
      "loss": 139.5692,
      "step": 8160
    },
    {
      "epoch": 0.06758489473466518,
      "grad_norm": 1140.02392578125,
      "learning_rate": 6.602020202020203e-06,
      "loss": 195.1031,
      "step": 8170
    },
    {
      "epoch": 0.06766761798403441,
      "grad_norm": 1378.050537109375,
      "learning_rate": 6.610101010101011e-06,
      "loss": 215.5058,
      "step": 8180
    },
    {
      "epoch": 0.06775034123340365,
      "grad_norm": 1974.1138916015625,
      "learning_rate": 6.618181818181819e-06,
      "loss": 294.2934,
      "step": 8190
    },
    {
      "epoch": 0.06783306448277289,
      "grad_norm": 832.8023681640625,
      "learning_rate": 6.626262626262627e-06,
      "loss": 184.4631,
      "step": 8200
    },
    {
      "epoch": 0.06791578773214212,
      "grad_norm": 856.9111938476562,
      "learning_rate": 6.634343434343435e-06,
      "loss": 170.4937,
      "step": 8210
    },
    {
      "epoch": 0.06799851098151136,
      "grad_norm": 916.113525390625,
      "learning_rate": 6.642424242424242e-06,
      "loss": 170.3063,
      "step": 8220
    },
    {
      "epoch": 0.06808123423088058,
      "grad_norm": 1267.204345703125,
      "learning_rate": 6.650505050505051e-06,
      "loss": 154.5532,
      "step": 8230
    },
    {
      "epoch": 0.06816395748024982,
      "grad_norm": 1163.3558349609375,
      "learning_rate": 6.658585858585859e-06,
      "loss": 196.9229,
      "step": 8240
    },
    {
      "epoch": 0.06824668072961906,
      "grad_norm": 1783.79541015625,
      "learning_rate": 6.666666666666667e-06,
      "loss": 210.5392,
      "step": 8250
    },
    {
      "epoch": 0.06832940397898829,
      "grad_norm": 917.7781982421875,
      "learning_rate": 6.674747474747475e-06,
      "loss": 177.4379,
      "step": 8260
    },
    {
      "epoch": 0.06841212722835753,
      "grad_norm": 674.7391967773438,
      "learning_rate": 6.682828282828283e-06,
      "loss": 186.2342,
      "step": 8270
    },
    {
      "epoch": 0.06849485047772677,
      "grad_norm": 654.15234375,
      "learning_rate": 6.690909090909091e-06,
      "loss": 136.4612,
      "step": 8280
    },
    {
      "epoch": 0.068577573727096,
      "grad_norm": 1094.0560302734375,
      "learning_rate": 6.6989898989899e-06,
      "loss": 170.5708,
      "step": 8290
    },
    {
      "epoch": 0.06866029697646524,
      "grad_norm": 1474.585693359375,
      "learning_rate": 6.707070707070707e-06,
      "loss": 203.4748,
      "step": 8300
    },
    {
      "epoch": 0.06874302022583446,
      "grad_norm": 1227.391845703125,
      "learning_rate": 6.715151515151516e-06,
      "loss": 222.0174,
      "step": 8310
    },
    {
      "epoch": 0.0688257434752037,
      "grad_norm": 1838.2691650390625,
      "learning_rate": 6.723232323232324e-06,
      "loss": 224.6444,
      "step": 8320
    },
    {
      "epoch": 0.06890846672457294,
      "grad_norm": 1169.060791015625,
      "learning_rate": 6.731313131313132e-06,
      "loss": 142.1164,
      "step": 8330
    },
    {
      "epoch": 0.06899118997394217,
      "grad_norm": 547.6967163085938,
      "learning_rate": 6.73939393939394e-06,
      "loss": 153.1282,
      "step": 8340
    },
    {
      "epoch": 0.06907391322331141,
      "grad_norm": 1083.063720703125,
      "learning_rate": 6.747474747474749e-06,
      "loss": 184.0296,
      "step": 8350
    },
    {
      "epoch": 0.06915663647268065,
      "grad_norm": 5596.943359375,
      "learning_rate": 6.755555555555556e-06,
      "loss": 173.6666,
      "step": 8360
    },
    {
      "epoch": 0.06923935972204988,
      "grad_norm": 1326.8721923828125,
      "learning_rate": 6.763636363636365e-06,
      "loss": 190.1673,
      "step": 8370
    },
    {
      "epoch": 0.06932208297141912,
      "grad_norm": 1101.3780517578125,
      "learning_rate": 6.771717171717172e-06,
      "loss": 220.6135,
      "step": 8380
    },
    {
      "epoch": 0.06940480622078836,
      "grad_norm": 1435.279296875,
      "learning_rate": 6.779797979797981e-06,
      "loss": 178.5454,
      "step": 8390
    },
    {
      "epoch": 0.06948752947015759,
      "grad_norm": 1620.093994140625,
      "learning_rate": 6.787878787878789e-06,
      "loss": 151.146,
      "step": 8400
    },
    {
      "epoch": 0.06957025271952683,
      "grad_norm": 1524.4627685546875,
      "learning_rate": 6.795959595959597e-06,
      "loss": 185.1478,
      "step": 8410
    },
    {
      "epoch": 0.06965297596889605,
      "grad_norm": 1806.7978515625,
      "learning_rate": 6.804040404040405e-06,
      "loss": 173.0628,
      "step": 8420
    },
    {
      "epoch": 0.06973569921826529,
      "grad_norm": 1826.8575439453125,
      "learning_rate": 6.812121212121212e-06,
      "loss": 167.8523,
      "step": 8430
    },
    {
      "epoch": 0.06981842246763453,
      "grad_norm": 1353.9345703125,
      "learning_rate": 6.82020202020202e-06,
      "loss": 194.5669,
      "step": 8440
    },
    {
      "epoch": 0.06990114571700376,
      "grad_norm": 2063.500244140625,
      "learning_rate": 6.828282828282828e-06,
      "loss": 184.4123,
      "step": 8450
    },
    {
      "epoch": 0.069983868966373,
      "grad_norm": 1244.160400390625,
      "learning_rate": 6.8363636363636364e-06,
      "loss": 234.5416,
      "step": 8460
    },
    {
      "epoch": 0.07006659221574224,
      "grad_norm": 894.0686645507812,
      "learning_rate": 6.844444444444445e-06,
      "loss": 157.8413,
      "step": 8470
    },
    {
      "epoch": 0.07014931546511147,
      "grad_norm": 953.1651000976562,
      "learning_rate": 6.8525252525252525e-06,
      "loss": 187.5382,
      "step": 8480
    },
    {
      "epoch": 0.07023203871448071,
      "grad_norm": 868.2008056640625,
      "learning_rate": 6.860606060606061e-06,
      "loss": 153.3808,
      "step": 8490
    },
    {
      "epoch": 0.07031476196384993,
      "grad_norm": 1246.61328125,
      "learning_rate": 6.868686868686869e-06,
      "loss": 198.8839,
      "step": 8500
    },
    {
      "epoch": 0.07039748521321917,
      "grad_norm": 1278.021484375,
      "learning_rate": 6.876767676767677e-06,
      "loss": 162.5859,
      "step": 8510
    },
    {
      "epoch": 0.07048020846258841,
      "grad_norm": 1224.3677978515625,
      "learning_rate": 6.8848484848484854e-06,
      "loss": 216.0751,
      "step": 8520
    },
    {
      "epoch": 0.07056293171195764,
      "grad_norm": 629.5072021484375,
      "learning_rate": 6.892929292929294e-06,
      "loss": 219.3646,
      "step": 8530
    },
    {
      "epoch": 0.07064565496132688,
      "grad_norm": 873.804443359375,
      "learning_rate": 6.9010101010101015e-06,
      "loss": 176.4796,
      "step": 8540
    },
    {
      "epoch": 0.07072837821069612,
      "grad_norm": 1832.682373046875,
      "learning_rate": 6.90909090909091e-06,
      "loss": 256.3848,
      "step": 8550
    },
    {
      "epoch": 0.07081110146006535,
      "grad_norm": 1244.7239990234375,
      "learning_rate": 6.9171717171717175e-06,
      "loss": 215.3179,
      "step": 8560
    },
    {
      "epoch": 0.07089382470943459,
      "grad_norm": 938.3501586914062,
      "learning_rate": 6.925252525252526e-06,
      "loss": 169.5242,
      "step": 8570
    },
    {
      "epoch": 0.07097654795880382,
      "grad_norm": 2076.439697265625,
      "learning_rate": 6.9333333333333344e-06,
      "loss": 200.3409,
      "step": 8580
    },
    {
      "epoch": 0.07105927120817306,
      "grad_norm": 800.4453735351562,
      "learning_rate": 6.941414141414142e-06,
      "loss": 192.4127,
      "step": 8590
    },
    {
      "epoch": 0.0711419944575423,
      "grad_norm": 1373.4234619140625,
      "learning_rate": 6.9494949494949505e-06,
      "loss": 161.1228,
      "step": 8600
    },
    {
      "epoch": 0.07122471770691152,
      "grad_norm": 950.77197265625,
      "learning_rate": 6.957575757575759e-06,
      "loss": 179.3362,
      "step": 8610
    },
    {
      "epoch": 0.07130744095628076,
      "grad_norm": 1121.46044921875,
      "learning_rate": 6.9656565656565665e-06,
      "loss": 196.0917,
      "step": 8620
    },
    {
      "epoch": 0.07139016420565,
      "grad_norm": 906.2406005859375,
      "learning_rate": 6.973737373737375e-06,
      "loss": 139.1432,
      "step": 8630
    },
    {
      "epoch": 0.07147288745501923,
      "grad_norm": 2017.980712890625,
      "learning_rate": 6.981818181818183e-06,
      "loss": 212.8915,
      "step": 8640
    },
    {
      "epoch": 0.07155561070438847,
      "grad_norm": 1261.973388671875,
      "learning_rate": 6.98989898989899e-06,
      "loss": 200.444,
      "step": 8650
    },
    {
      "epoch": 0.07163833395375771,
      "grad_norm": 855.2569580078125,
      "learning_rate": 6.997979797979798e-06,
      "loss": 222.0111,
      "step": 8660
    },
    {
      "epoch": 0.07172105720312694,
      "grad_norm": 1921.39599609375,
      "learning_rate": 7.006060606060606e-06,
      "loss": 222.5992,
      "step": 8670
    },
    {
      "epoch": 0.07180378045249618,
      "grad_norm": 857.9965209960938,
      "learning_rate": 7.014141414141415e-06,
      "loss": 219.586,
      "step": 8680
    },
    {
      "epoch": 0.0718865037018654,
      "grad_norm": 915.5559692382812,
      "learning_rate": 7.022222222222222e-06,
      "loss": 184.9689,
      "step": 8690
    },
    {
      "epoch": 0.07196922695123464,
      "grad_norm": 1838.1370849609375,
      "learning_rate": 7.030303030303031e-06,
      "loss": 232.5889,
      "step": 8700
    },
    {
      "epoch": 0.07205195020060388,
      "grad_norm": 932.5780639648438,
      "learning_rate": 7.038383838383839e-06,
      "loss": 171.6244,
      "step": 8710
    },
    {
      "epoch": 0.07213467344997311,
      "grad_norm": 1225.2701416015625,
      "learning_rate": 7.046464646464647e-06,
      "loss": 153.8994,
      "step": 8720
    },
    {
      "epoch": 0.07221739669934235,
      "grad_norm": 1276.4774169921875,
      "learning_rate": 7.054545454545455e-06,
      "loss": 195.4201,
      "step": 8730
    },
    {
      "epoch": 0.07230011994871159,
      "grad_norm": 1169.0623779296875,
      "learning_rate": 7.062626262626263e-06,
      "loss": 184.115,
      "step": 8740
    },
    {
      "epoch": 0.07238284319808082,
      "grad_norm": 1720.584716796875,
      "learning_rate": 7.070707070707071e-06,
      "loss": 225.3465,
      "step": 8750
    },
    {
      "epoch": 0.07246556644745006,
      "grad_norm": 811.720947265625,
      "learning_rate": 7.07878787878788e-06,
      "loss": 155.7371,
      "step": 8760
    },
    {
      "epoch": 0.07254828969681928,
      "grad_norm": 1744.4354248046875,
      "learning_rate": 7.086868686868687e-06,
      "loss": 193.3227,
      "step": 8770
    },
    {
      "epoch": 0.07263101294618853,
      "grad_norm": 862.0523681640625,
      "learning_rate": 7.094949494949496e-06,
      "loss": 170.9686,
      "step": 8780
    },
    {
      "epoch": 0.07271373619555777,
      "grad_norm": 2165.270751953125,
      "learning_rate": 7.103030303030304e-06,
      "loss": 234.9279,
      "step": 8790
    },
    {
      "epoch": 0.07279645944492699,
      "grad_norm": 1295.5478515625,
      "learning_rate": 7.111111111111112e-06,
      "loss": 211.1977,
      "step": 8800
    },
    {
      "epoch": 0.07287918269429623,
      "grad_norm": 957.7193603515625,
      "learning_rate": 7.11919191919192e-06,
      "loss": 191.1015,
      "step": 8810
    },
    {
      "epoch": 0.07296190594366547,
      "grad_norm": 1650.5340576171875,
      "learning_rate": 7.127272727272728e-06,
      "loss": 164.5978,
      "step": 8820
    },
    {
      "epoch": 0.0730446291930347,
      "grad_norm": 1488.514404296875,
      "learning_rate": 7.135353535353536e-06,
      "loss": 187.6664,
      "step": 8830
    },
    {
      "epoch": 0.07312735244240394,
      "grad_norm": 1756.99853515625,
      "learning_rate": 7.143434343434345e-06,
      "loss": 219.4122,
      "step": 8840
    },
    {
      "epoch": 0.07321007569177317,
      "grad_norm": 924.37158203125,
      "learning_rate": 7.151515151515152e-06,
      "loss": 157.6518,
      "step": 8850
    },
    {
      "epoch": 0.0732927989411424,
      "grad_norm": 1157.555908203125,
      "learning_rate": 7.15959595959596e-06,
      "loss": 162.5851,
      "step": 8860
    },
    {
      "epoch": 0.07337552219051165,
      "grad_norm": 668.1375732421875,
      "learning_rate": 7.1676767676767676e-06,
      "loss": 153.8082,
      "step": 8870
    },
    {
      "epoch": 0.07345824543988087,
      "grad_norm": 1706.42333984375,
      "learning_rate": 7.175757575757576e-06,
      "loss": 145.3626,
      "step": 8880
    },
    {
      "epoch": 0.07354096868925011,
      "grad_norm": 2563.07861328125,
      "learning_rate": 7.1838383838383845e-06,
      "loss": 217.706,
      "step": 8890
    },
    {
      "epoch": 0.07362369193861935,
      "grad_norm": 1756.02392578125,
      "learning_rate": 7.191919191919192e-06,
      "loss": 177.5173,
      "step": 8900
    },
    {
      "epoch": 0.07370641518798858,
      "grad_norm": 1470.858642578125,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 218.2071,
      "step": 8910
    },
    {
      "epoch": 0.07378913843735782,
      "grad_norm": 1429.1534423828125,
      "learning_rate": 7.208080808080808e-06,
      "loss": 192.6579,
      "step": 8920
    },
    {
      "epoch": 0.07387186168672706,
      "grad_norm": 1044.5311279296875,
      "learning_rate": 7.2161616161616166e-06,
      "loss": 175.2781,
      "step": 8930
    },
    {
      "epoch": 0.07395458493609629,
      "grad_norm": 1081.16748046875,
      "learning_rate": 7.224242424242425e-06,
      "loss": 195.337,
      "step": 8940
    },
    {
      "epoch": 0.07403730818546553,
      "grad_norm": 1816.50634765625,
      "learning_rate": 7.232323232323233e-06,
      "loss": 164.159,
      "step": 8950
    },
    {
      "epoch": 0.07412003143483475,
      "grad_norm": 941.095947265625,
      "learning_rate": 7.240404040404041e-06,
      "loss": 167.4316,
      "step": 8960
    },
    {
      "epoch": 0.074202754684204,
      "grad_norm": 1430.9918212890625,
      "learning_rate": 7.2484848484848495e-06,
      "loss": 142.7207,
      "step": 8970
    },
    {
      "epoch": 0.07428547793357324,
      "grad_norm": 987.8670043945312,
      "learning_rate": 7.256565656565657e-06,
      "loss": 185.2262,
      "step": 8980
    },
    {
      "epoch": 0.07436820118294246,
      "grad_norm": 1886.202392578125,
      "learning_rate": 7.2646464646464656e-06,
      "loss": 201.7182,
      "step": 8990
    },
    {
      "epoch": 0.0744509244323117,
      "grad_norm": 6469.64501953125,
      "learning_rate": 7.272727272727273e-06,
      "loss": 242.3246,
      "step": 9000
    },
    {
      "epoch": 0.07453364768168094,
      "grad_norm": 1446.23388671875,
      "learning_rate": 7.280808080808082e-06,
      "loss": 185.7719,
      "step": 9010
    },
    {
      "epoch": 0.07461637093105017,
      "grad_norm": 1158.1019287109375,
      "learning_rate": 7.28888888888889e-06,
      "loss": 135.7039,
      "step": 9020
    },
    {
      "epoch": 0.07469909418041941,
      "grad_norm": 1710.8179931640625,
      "learning_rate": 7.296969696969698e-06,
      "loss": 214.3662,
      "step": 9030
    },
    {
      "epoch": 0.07478181742978864,
      "grad_norm": 693.0316772460938,
      "learning_rate": 7.305050505050506e-06,
      "loss": 168.4758,
      "step": 9040
    },
    {
      "epoch": 0.07486454067915788,
      "grad_norm": 1307.99267578125,
      "learning_rate": 7.3131313131313146e-06,
      "loss": 221.4772,
      "step": 9050
    },
    {
      "epoch": 0.07494726392852712,
      "grad_norm": 578.5460815429688,
      "learning_rate": 7.321212121212122e-06,
      "loss": 172.2857,
      "step": 9060
    },
    {
      "epoch": 0.07502998717789634,
      "grad_norm": 4154.9052734375,
      "learning_rate": 7.32929292929293e-06,
      "loss": 200.6511,
      "step": 9070
    },
    {
      "epoch": 0.07511271042726558,
      "grad_norm": 3818.542236328125,
      "learning_rate": 7.337373737373737e-06,
      "loss": 198.6826,
      "step": 9080
    },
    {
      "epoch": 0.07519543367663482,
      "grad_norm": 630.0006713867188,
      "learning_rate": 7.345454545454546e-06,
      "loss": 146.3737,
      "step": 9090
    },
    {
      "epoch": 0.07527815692600405,
      "grad_norm": 1059.06640625,
      "learning_rate": 7.353535353535353e-06,
      "loss": 163.5028,
      "step": 9100
    },
    {
      "epoch": 0.07536088017537329,
      "grad_norm": 668.64111328125,
      "learning_rate": 7.361616161616162e-06,
      "loss": 170.1274,
      "step": 9110
    },
    {
      "epoch": 0.07544360342474253,
      "grad_norm": 1227.2437744140625,
      "learning_rate": 7.36969696969697e-06,
      "loss": 167.1523,
      "step": 9120
    },
    {
      "epoch": 0.07552632667411176,
      "grad_norm": 989.3958740234375,
      "learning_rate": 7.377777777777778e-06,
      "loss": 179.0444,
      "step": 9130
    },
    {
      "epoch": 0.075609049923481,
      "grad_norm": 2812.70166015625,
      "learning_rate": 7.385858585858586e-06,
      "loss": 170.3505,
      "step": 9140
    },
    {
      "epoch": 0.07569177317285022,
      "grad_norm": 1429.5191650390625,
      "learning_rate": 7.393939393939395e-06,
      "loss": 198.7005,
      "step": 9150
    },
    {
      "epoch": 0.07577449642221946,
      "grad_norm": 937.4029541015625,
      "learning_rate": 7.402020202020202e-06,
      "loss": 149.7339,
      "step": 9160
    },
    {
      "epoch": 0.0758572196715887,
      "grad_norm": 6428.83642578125,
      "learning_rate": 7.410101010101011e-06,
      "loss": 224.5054,
      "step": 9170
    },
    {
      "epoch": 0.07593994292095793,
      "grad_norm": 1135.309814453125,
      "learning_rate": 7.4181818181818185e-06,
      "loss": 206.6003,
      "step": 9180
    },
    {
      "epoch": 0.07602266617032717,
      "grad_norm": 663.953125,
      "learning_rate": 7.426262626262627e-06,
      "loss": 148.6399,
      "step": 9190
    },
    {
      "epoch": 0.07610538941969641,
      "grad_norm": 1465.2401123046875,
      "learning_rate": 7.434343434343435e-06,
      "loss": 129.7046,
      "step": 9200
    },
    {
      "epoch": 0.07618811266906564,
      "grad_norm": 840.3555297851562,
      "learning_rate": 7.442424242424243e-06,
      "loss": 212.8563,
      "step": 9210
    },
    {
      "epoch": 0.07627083591843488,
      "grad_norm": 1056.4595947265625,
      "learning_rate": 7.450505050505051e-06,
      "loss": 219.987,
      "step": 9220
    },
    {
      "epoch": 0.0763535591678041,
      "grad_norm": 1585.2525634765625,
      "learning_rate": 7.45858585858586e-06,
      "loss": 254.3914,
      "step": 9230
    },
    {
      "epoch": 0.07643628241717335,
      "grad_norm": 1167.71728515625,
      "learning_rate": 7.4666666666666675e-06,
      "loss": 173.4369,
      "step": 9240
    },
    {
      "epoch": 0.07651900566654259,
      "grad_norm": 1241.547119140625,
      "learning_rate": 7.474747474747476e-06,
      "loss": 196.5241,
      "step": 9250
    },
    {
      "epoch": 0.07660172891591181,
      "grad_norm": 1076.36279296875,
      "learning_rate": 7.4828282828282835e-06,
      "loss": 141.0298,
      "step": 9260
    },
    {
      "epoch": 0.07668445216528105,
      "grad_norm": 894.4515991210938,
      "learning_rate": 7.490909090909092e-06,
      "loss": 182.4499,
      "step": 9270
    },
    {
      "epoch": 0.0767671754146503,
      "grad_norm": 1318.6171875,
      "learning_rate": 7.4989898989899e-06,
      "loss": 294.3835,
      "step": 9280
    },
    {
      "epoch": 0.07684989866401952,
      "grad_norm": 1835.936279296875,
      "learning_rate": 7.507070707070707e-06,
      "loss": 200.7481,
      "step": 9290
    },
    {
      "epoch": 0.07693262191338876,
      "grad_norm": 1094.5806884765625,
      "learning_rate": 7.515151515151516e-06,
      "loss": 155.6739,
      "step": 9300
    },
    {
      "epoch": 0.07701534516275799,
      "grad_norm": 2438.4599609375,
      "learning_rate": 7.523232323232323e-06,
      "loss": 168.196,
      "step": 9310
    },
    {
      "epoch": 0.07709806841212723,
      "grad_norm": 1409.369384765625,
      "learning_rate": 7.531313131313132e-06,
      "loss": 168.5339,
      "step": 9320
    },
    {
      "epoch": 0.07718079166149647,
      "grad_norm": 792.833740234375,
      "learning_rate": 7.53939393939394e-06,
      "loss": 204.8236,
      "step": 9330
    },
    {
      "epoch": 0.0772635149108657,
      "grad_norm": 4106.85107421875,
      "learning_rate": 7.547474747474748e-06,
      "loss": 206.0253,
      "step": 9340
    },
    {
      "epoch": 0.07734623816023493,
      "grad_norm": 1158.9046630859375,
      "learning_rate": 7.555555555555556e-06,
      "loss": 200.0595,
      "step": 9350
    },
    {
      "epoch": 0.07742896140960417,
      "grad_norm": 829.5303344726562,
      "learning_rate": 7.563636363636364e-06,
      "loss": 152.1526,
      "step": 9360
    },
    {
      "epoch": 0.0775116846589734,
      "grad_norm": 1672.3824462890625,
      "learning_rate": 7.571717171717172e-06,
      "loss": 196.9728,
      "step": 9370
    },
    {
      "epoch": 0.07759440790834264,
      "grad_norm": 1697.9437255859375,
      "learning_rate": 7.579797979797981e-06,
      "loss": 140.7818,
      "step": 9380
    },
    {
      "epoch": 0.07767713115771188,
      "grad_norm": 1467.42138671875,
      "learning_rate": 7.587878787878788e-06,
      "loss": 237.8365,
      "step": 9390
    },
    {
      "epoch": 0.07775985440708111,
      "grad_norm": 1372.5516357421875,
      "learning_rate": 7.595959595959597e-06,
      "loss": 201.537,
      "step": 9400
    },
    {
      "epoch": 0.07784257765645035,
      "grad_norm": 1280.147216796875,
      "learning_rate": 7.604040404040405e-06,
      "loss": 168.6732,
      "step": 9410
    },
    {
      "epoch": 0.07792530090581958,
      "grad_norm": 1045.321533203125,
      "learning_rate": 7.612121212121213e-06,
      "loss": 154.092,
      "step": 9420
    },
    {
      "epoch": 0.07800802415518882,
      "grad_norm": 6423.11962890625,
      "learning_rate": 7.620202020202021e-06,
      "loss": 251.8613,
      "step": 9430
    },
    {
      "epoch": 0.07809074740455806,
      "grad_norm": 1650.4332275390625,
      "learning_rate": 7.628282828282829e-06,
      "loss": 218.2032,
      "step": 9440
    },
    {
      "epoch": 0.07817347065392728,
      "grad_norm": 1520.5958251953125,
      "learning_rate": 7.636363636363638e-06,
      "loss": 194.5603,
      "step": 9450
    },
    {
      "epoch": 0.07825619390329652,
      "grad_norm": 1989.30126953125,
      "learning_rate": 7.644444444444445e-06,
      "loss": 167.2613,
      "step": 9460
    },
    {
      "epoch": 0.07833891715266576,
      "grad_norm": 1363.1075439453125,
      "learning_rate": 7.652525252525253e-06,
      "loss": 193.0324,
      "step": 9470
    },
    {
      "epoch": 0.07842164040203499,
      "grad_norm": 1044.422607421875,
      "learning_rate": 7.660606060606062e-06,
      "loss": 239.5535,
      "step": 9480
    },
    {
      "epoch": 0.07850436365140423,
      "grad_norm": 1506.1378173828125,
      "learning_rate": 7.66868686868687e-06,
      "loss": 157.3207,
      "step": 9490
    },
    {
      "epoch": 0.07858708690077346,
      "grad_norm": 1023.1807250976562,
      "learning_rate": 7.676767676767677e-06,
      "loss": 177.0086,
      "step": 9500
    },
    {
      "epoch": 0.0786698101501427,
      "grad_norm": 766.3860473632812,
      "learning_rate": 7.684848484848485e-06,
      "loss": 180.8027,
      "step": 9510
    },
    {
      "epoch": 0.07875253339951194,
      "grad_norm": 1220.765380859375,
      "learning_rate": 7.692929292929294e-06,
      "loss": 199.764,
      "step": 9520
    },
    {
      "epoch": 0.07883525664888116,
      "grad_norm": 798.783447265625,
      "learning_rate": 7.7010101010101e-06,
      "loss": 178.7332,
      "step": 9530
    },
    {
      "epoch": 0.0789179798982504,
      "grad_norm": 1135.2913818359375,
      "learning_rate": 7.709090909090909e-06,
      "loss": 138.7885,
      "step": 9540
    },
    {
      "epoch": 0.07900070314761964,
      "grad_norm": 820.0831298828125,
      "learning_rate": 7.717171717171717e-06,
      "loss": 192.7479,
      "step": 9550
    },
    {
      "epoch": 0.07908342639698887,
      "grad_norm": 1487.1383056640625,
      "learning_rate": 7.725252525252526e-06,
      "loss": 202.4578,
      "step": 9560
    },
    {
      "epoch": 0.07916614964635811,
      "grad_norm": 3294.249267578125,
      "learning_rate": 7.733333333333334e-06,
      "loss": 154.1931,
      "step": 9570
    },
    {
      "epoch": 0.07924887289572734,
      "grad_norm": 1593.376708984375,
      "learning_rate": 7.741414141414141e-06,
      "loss": 220.1035,
      "step": 9580
    },
    {
      "epoch": 0.07933159614509658,
      "grad_norm": 1254.923095703125,
      "learning_rate": 7.74949494949495e-06,
      "loss": 162.3905,
      "step": 9590
    },
    {
      "epoch": 0.07941431939446582,
      "grad_norm": 989.010009765625,
      "learning_rate": 7.757575757575758e-06,
      "loss": 150.6632,
      "step": 9600
    },
    {
      "epoch": 0.07949704264383504,
      "grad_norm": 668.603759765625,
      "learning_rate": 7.765656565656566e-06,
      "loss": 180.0927,
      "step": 9610
    },
    {
      "epoch": 0.07957976589320429,
      "grad_norm": 918.044677734375,
      "learning_rate": 7.773737373737375e-06,
      "loss": 216.1173,
      "step": 9620
    },
    {
      "epoch": 0.07966248914257353,
      "grad_norm": 1150.803955078125,
      "learning_rate": 7.781818181818183e-06,
      "loss": 165.5561,
      "step": 9630
    },
    {
      "epoch": 0.07974521239194275,
      "grad_norm": 1339.1085205078125,
      "learning_rate": 7.78989898989899e-06,
      "loss": 120.4266,
      "step": 9640
    },
    {
      "epoch": 0.07982793564131199,
      "grad_norm": 1119.98388671875,
      "learning_rate": 7.797979797979799e-06,
      "loss": 186.139,
      "step": 9650
    },
    {
      "epoch": 0.07991065889068123,
      "grad_norm": 845.5322875976562,
      "learning_rate": 7.806060606060607e-06,
      "loss": 147.2671,
      "step": 9660
    },
    {
      "epoch": 0.07999338214005046,
      "grad_norm": 1525.1080322265625,
      "learning_rate": 7.814141414141415e-06,
      "loss": 227.4823,
      "step": 9670
    },
    {
      "epoch": 0.0800761053894197,
      "grad_norm": 936.9368286132812,
      "learning_rate": 7.822222222222224e-06,
      "loss": 147.8492,
      "step": 9680
    },
    {
      "epoch": 0.08015882863878893,
      "grad_norm": 1528.34375,
      "learning_rate": 7.83030303030303e-06,
      "loss": 176.4171,
      "step": 9690
    },
    {
      "epoch": 0.08024155188815817,
      "grad_norm": 1028.9813232421875,
      "learning_rate": 7.838383838383839e-06,
      "loss": 168.1568,
      "step": 9700
    },
    {
      "epoch": 0.0803242751375274,
      "grad_norm": 3313.676513671875,
      "learning_rate": 7.846464646464646e-06,
      "loss": 211.1931,
      "step": 9710
    },
    {
      "epoch": 0.08040699838689663,
      "grad_norm": 698.055419921875,
      "learning_rate": 7.854545454545454e-06,
      "loss": 147.0731,
      "step": 9720
    },
    {
      "epoch": 0.08048972163626587,
      "grad_norm": 1256.8193359375,
      "learning_rate": 7.862626262626263e-06,
      "loss": 141.3686,
      "step": 9730
    },
    {
      "epoch": 0.08057244488563511,
      "grad_norm": 1118.2659912109375,
      "learning_rate": 7.870707070707071e-06,
      "loss": 191.6792,
      "step": 9740
    },
    {
      "epoch": 0.08065516813500434,
      "grad_norm": 1615.0177001953125,
      "learning_rate": 7.87878787878788e-06,
      "loss": 171.2617,
      "step": 9750
    },
    {
      "epoch": 0.08073789138437358,
      "grad_norm": 1359.4404296875,
      "learning_rate": 7.886868686868686e-06,
      "loss": 233.8179,
      "step": 9760
    },
    {
      "epoch": 0.08082061463374281,
      "grad_norm": 1080.4814453125,
      "learning_rate": 7.894949494949495e-06,
      "loss": 167.5727,
      "step": 9770
    },
    {
      "epoch": 0.08090333788311205,
      "grad_norm": 1568.5997314453125,
      "learning_rate": 7.903030303030303e-06,
      "loss": 180.408,
      "step": 9780
    },
    {
      "epoch": 0.08098606113248129,
      "grad_norm": 1154.7166748046875,
      "learning_rate": 7.911111111111112e-06,
      "loss": 141.3988,
      "step": 9790
    },
    {
      "epoch": 0.08106878438185051,
      "grad_norm": 1465.5701904296875,
      "learning_rate": 7.91919191919192e-06,
      "loss": 235.8258,
      "step": 9800
    },
    {
      "epoch": 0.08115150763121975,
      "grad_norm": 1399.992919921875,
      "learning_rate": 7.927272727272729e-06,
      "loss": 197.8094,
      "step": 9810
    },
    {
      "epoch": 0.081234230880589,
      "grad_norm": 1609.233154296875,
      "learning_rate": 7.935353535353535e-06,
      "loss": 169.6143,
      "step": 9820
    },
    {
      "epoch": 0.08131695412995822,
      "grad_norm": 771.2155151367188,
      "learning_rate": 7.943434343434344e-06,
      "loss": 174.0228,
      "step": 9830
    },
    {
      "epoch": 0.08139967737932746,
      "grad_norm": 1397.8433837890625,
      "learning_rate": 7.951515151515152e-06,
      "loss": 195.8236,
      "step": 9840
    },
    {
      "epoch": 0.08148240062869669,
      "grad_norm": 1942.0560302734375,
      "learning_rate": 7.95959595959596e-06,
      "loss": 174.9514,
      "step": 9850
    },
    {
      "epoch": 0.08156512387806593,
      "grad_norm": 881.597900390625,
      "learning_rate": 7.96767676767677e-06,
      "loss": 155.3664,
      "step": 9860
    },
    {
      "epoch": 0.08164784712743517,
      "grad_norm": 982.1563110351562,
      "learning_rate": 7.975757575757576e-06,
      "loss": 166.2256,
      "step": 9870
    },
    {
      "epoch": 0.0817305703768044,
      "grad_norm": 1786.5938720703125,
      "learning_rate": 7.983838383838384e-06,
      "loss": 179.014,
      "step": 9880
    },
    {
      "epoch": 0.08181329362617364,
      "grad_norm": 1422.998046875,
      "learning_rate": 7.991919191919193e-06,
      "loss": 185.8182,
      "step": 9890
    },
    {
      "epoch": 0.08189601687554288,
      "grad_norm": 1009.948486328125,
      "learning_rate": 8.000000000000001e-06,
      "loss": 153.7416,
      "step": 9900
    },
    {
      "epoch": 0.0819787401249121,
      "grad_norm": 929.018798828125,
      "learning_rate": 8.00808080808081e-06,
      "loss": 178.9259,
      "step": 9910
    },
    {
      "epoch": 0.08206146337428134,
      "grad_norm": 654.3829345703125,
      "learning_rate": 8.016161616161617e-06,
      "loss": 130.9901,
      "step": 9920
    },
    {
      "epoch": 0.08214418662365058,
      "grad_norm": 1169.20068359375,
      "learning_rate": 8.024242424242425e-06,
      "loss": 125.0886,
      "step": 9930
    },
    {
      "epoch": 0.08222690987301981,
      "grad_norm": 1203.2630615234375,
      "learning_rate": 8.032323232323232e-06,
      "loss": 191.309,
      "step": 9940
    },
    {
      "epoch": 0.08230963312238905,
      "grad_norm": 1353.7027587890625,
      "learning_rate": 8.04040404040404e-06,
      "loss": 239.4487,
      "step": 9950
    },
    {
      "epoch": 0.08239235637175828,
      "grad_norm": 1199.4378662109375,
      "learning_rate": 8.048484848484849e-06,
      "loss": 173.5184,
      "step": 9960
    },
    {
      "epoch": 0.08247507962112752,
      "grad_norm": 1027.1614990234375,
      "learning_rate": 8.056565656565657e-06,
      "loss": 236.1041,
      "step": 9970
    },
    {
      "epoch": 0.08255780287049676,
      "grad_norm": 2071.078125,
      "learning_rate": 8.064646464646466e-06,
      "loss": 210.167,
      "step": 9980
    },
    {
      "epoch": 0.08264052611986598,
      "grad_norm": 931.7273559570312,
      "learning_rate": 8.072727272727274e-06,
      "loss": 169.0378,
      "step": 9990
    },
    {
      "epoch": 0.08272324936923522,
      "grad_norm": 1780.4796142578125,
      "learning_rate": 8.08080808080808e-06,
      "loss": 182.981,
      "step": 10000
    },
    {
      "epoch": 0.08280597261860446,
      "grad_norm": 1484.16845703125,
      "learning_rate": 8.08888888888889e-06,
      "loss": 190.9732,
      "step": 10010
    },
    {
      "epoch": 0.08288869586797369,
      "grad_norm": 1517.7740478515625,
      "learning_rate": 8.096969696969698e-06,
      "loss": 214.248,
      "step": 10020
    },
    {
      "epoch": 0.08297141911734293,
      "grad_norm": 4507.2001953125,
      "learning_rate": 8.105050505050506e-06,
      "loss": 149.1075,
      "step": 10030
    },
    {
      "epoch": 0.08305414236671216,
      "grad_norm": 1202.4232177734375,
      "learning_rate": 8.113131313131315e-06,
      "loss": 167.8956,
      "step": 10040
    },
    {
      "epoch": 0.0831368656160814,
      "grad_norm": 3145.95556640625,
      "learning_rate": 8.121212121212121e-06,
      "loss": 218.0878,
      "step": 10050
    },
    {
      "epoch": 0.08321958886545064,
      "grad_norm": 833.5780029296875,
      "learning_rate": 8.12929292929293e-06,
      "loss": 183.5963,
      "step": 10060
    },
    {
      "epoch": 0.08330231211481987,
      "grad_norm": 1200.9332275390625,
      "learning_rate": 8.137373737373738e-06,
      "loss": 200.6742,
      "step": 10070
    },
    {
      "epoch": 0.0833850353641891,
      "grad_norm": 2536.378662109375,
      "learning_rate": 8.145454545454547e-06,
      "loss": 244.585,
      "step": 10080
    },
    {
      "epoch": 0.08346775861355835,
      "grad_norm": 0.0,
      "learning_rate": 8.153535353535355e-06,
      "loss": 151.4828,
      "step": 10090
    },
    {
      "epoch": 0.08355048186292757,
      "grad_norm": 1108.849853515625,
      "learning_rate": 8.161616161616162e-06,
      "loss": 178.3399,
      "step": 10100
    },
    {
      "epoch": 0.08363320511229681,
      "grad_norm": 809.9747314453125,
      "learning_rate": 8.16969696969697e-06,
      "loss": 201.301,
      "step": 10110
    },
    {
      "epoch": 0.08371592836166604,
      "grad_norm": 1331.628173828125,
      "learning_rate": 8.177777777777779e-06,
      "loss": 141.3669,
      "step": 10120
    },
    {
      "epoch": 0.08379865161103528,
      "grad_norm": 990.9824829101562,
      "learning_rate": 8.185858585858587e-06,
      "loss": 169.3207,
      "step": 10130
    },
    {
      "epoch": 0.08388137486040452,
      "grad_norm": 1191.9073486328125,
      "learning_rate": 8.193939393939394e-06,
      "loss": 136.0156,
      "step": 10140
    },
    {
      "epoch": 0.08396409810977375,
      "grad_norm": 829.3964233398438,
      "learning_rate": 8.202020202020202e-06,
      "loss": 165.8068,
      "step": 10150
    },
    {
      "epoch": 0.08404682135914299,
      "grad_norm": 781.4263305664062,
      "learning_rate": 8.21010101010101e-06,
      "loss": 186.8697,
      "step": 10160
    },
    {
      "epoch": 0.08412954460851223,
      "grad_norm": 815.2606811523438,
      "learning_rate": 8.21818181818182e-06,
      "loss": 257.8631,
      "step": 10170
    },
    {
      "epoch": 0.08421226785788145,
      "grad_norm": 1860.0218505859375,
      "learning_rate": 8.226262626262626e-06,
      "loss": 187.696,
      "step": 10180
    },
    {
      "epoch": 0.0842949911072507,
      "grad_norm": 2099.6728515625,
      "learning_rate": 8.234343434343434e-06,
      "loss": 205.9569,
      "step": 10190
    },
    {
      "epoch": 0.08437771435661993,
      "grad_norm": 1526.5322265625,
      "learning_rate": 8.242424242424243e-06,
      "loss": 200.3001,
      "step": 10200
    },
    {
      "epoch": 0.08446043760598916,
      "grad_norm": 1122.8441162109375,
      "learning_rate": 8.250505050505051e-06,
      "loss": 132.1188,
      "step": 10210
    },
    {
      "epoch": 0.0845431608553584,
      "grad_norm": 952.8866577148438,
      "learning_rate": 8.25858585858586e-06,
      "loss": 180.2814,
      "step": 10220
    },
    {
      "epoch": 0.08462588410472763,
      "grad_norm": 1098.6842041015625,
      "learning_rate": 8.266666666666667e-06,
      "loss": 186.863,
      "step": 10230
    },
    {
      "epoch": 0.08470860735409687,
      "grad_norm": 1201.576904296875,
      "learning_rate": 8.274747474747475e-06,
      "loss": 166.658,
      "step": 10240
    },
    {
      "epoch": 0.08479133060346611,
      "grad_norm": 3628.583984375,
      "learning_rate": 8.282828282828283e-06,
      "loss": 280.9522,
      "step": 10250
    },
    {
      "epoch": 0.08487405385283533,
      "grad_norm": 956.4686279296875,
      "learning_rate": 8.290909090909092e-06,
      "loss": 177.9082,
      "step": 10260
    },
    {
      "epoch": 0.08495677710220458,
      "grad_norm": 1213.8653564453125,
      "learning_rate": 8.2989898989899e-06,
      "loss": 176.0302,
      "step": 10270
    },
    {
      "epoch": 0.08503950035157382,
      "grad_norm": 1263.095458984375,
      "learning_rate": 8.307070707070707e-06,
      "loss": 171.6865,
      "step": 10280
    },
    {
      "epoch": 0.08512222360094304,
      "grad_norm": 1094.4754638671875,
      "learning_rate": 8.315151515151516e-06,
      "loss": 158.2062,
      "step": 10290
    },
    {
      "epoch": 0.08520494685031228,
      "grad_norm": 957.9495239257812,
      "learning_rate": 8.323232323232324e-06,
      "loss": 169.5506,
      "step": 10300
    },
    {
      "epoch": 0.08528767009968151,
      "grad_norm": 1073.7796630859375,
      "learning_rate": 8.331313131313132e-06,
      "loss": 147.6098,
      "step": 10310
    },
    {
      "epoch": 0.08537039334905075,
      "grad_norm": 1453.0513916015625,
      "learning_rate": 8.339393939393941e-06,
      "loss": 189.7265,
      "step": 10320
    },
    {
      "epoch": 0.08545311659841999,
      "grad_norm": 791.720703125,
      "learning_rate": 8.34747474747475e-06,
      "loss": 214.0726,
      "step": 10330
    },
    {
      "epoch": 0.08553583984778922,
      "grad_norm": 1202.690673828125,
      "learning_rate": 8.355555555555556e-06,
      "loss": 197.4867,
      "step": 10340
    },
    {
      "epoch": 0.08561856309715846,
      "grad_norm": 1175.0545654296875,
      "learning_rate": 8.363636363636365e-06,
      "loss": 153.0685,
      "step": 10350
    },
    {
      "epoch": 0.0857012863465277,
      "grad_norm": 745.9451293945312,
      "learning_rate": 8.371717171717171e-06,
      "loss": 170.8535,
      "step": 10360
    },
    {
      "epoch": 0.08578400959589692,
      "grad_norm": 1357.7061767578125,
      "learning_rate": 8.37979797979798e-06,
      "loss": 162.8026,
      "step": 10370
    },
    {
      "epoch": 0.08586673284526616,
      "grad_norm": 1293.4013671875,
      "learning_rate": 8.387878787878788e-06,
      "loss": 152.5479,
      "step": 10380
    },
    {
      "epoch": 0.0859494560946354,
      "grad_norm": 1522.9151611328125,
      "learning_rate": 8.395959595959597e-06,
      "loss": 180.1763,
      "step": 10390
    },
    {
      "epoch": 0.08603217934400463,
      "grad_norm": 2097.494140625,
      "learning_rate": 8.404040404040405e-06,
      "loss": 168.4439,
      "step": 10400
    },
    {
      "epoch": 0.08611490259337387,
      "grad_norm": 1112.0291748046875,
      "learning_rate": 8.412121212121212e-06,
      "loss": 193.0542,
      "step": 10410
    },
    {
      "epoch": 0.0861976258427431,
      "grad_norm": 927.9266357421875,
      "learning_rate": 8.42020202020202e-06,
      "loss": 211.676,
      "step": 10420
    },
    {
      "epoch": 0.08628034909211234,
      "grad_norm": 1505.004150390625,
      "learning_rate": 8.428282828282829e-06,
      "loss": 203.2798,
      "step": 10430
    },
    {
      "epoch": 0.08636307234148158,
      "grad_norm": 1962.483154296875,
      "learning_rate": 8.436363636363637e-06,
      "loss": 217.3728,
      "step": 10440
    },
    {
      "epoch": 0.0864457955908508,
      "grad_norm": 2074.048583984375,
      "learning_rate": 8.444444444444446e-06,
      "loss": 224.9844,
      "step": 10450
    },
    {
      "epoch": 0.08652851884022005,
      "grad_norm": 1316.5950927734375,
      "learning_rate": 8.452525252525252e-06,
      "loss": 165.5474,
      "step": 10460
    },
    {
      "epoch": 0.08661124208958929,
      "grad_norm": 1096.7896728515625,
      "learning_rate": 8.460606060606061e-06,
      "loss": 138.2374,
      "step": 10470
    },
    {
      "epoch": 0.08669396533895851,
      "grad_norm": 1451.124755859375,
      "learning_rate": 8.46868686868687e-06,
      "loss": 141.8237,
      "step": 10480
    },
    {
      "epoch": 0.08677668858832775,
      "grad_norm": 1239.64697265625,
      "learning_rate": 8.476767676767678e-06,
      "loss": 156.3808,
      "step": 10490
    },
    {
      "epoch": 0.08685941183769698,
      "grad_norm": 689.178955078125,
      "learning_rate": 8.484848484848486e-06,
      "loss": 143.1666,
      "step": 10500
    },
    {
      "epoch": 0.08694213508706622,
      "grad_norm": 684.1348266601562,
      "learning_rate": 8.492929292929295e-06,
      "loss": 182.6362,
      "step": 10510
    },
    {
      "epoch": 0.08702485833643546,
      "grad_norm": 1299.8443603515625,
      "learning_rate": 8.501010101010101e-06,
      "loss": 160.0256,
      "step": 10520
    },
    {
      "epoch": 0.08710758158580469,
      "grad_norm": 845.9354858398438,
      "learning_rate": 8.50909090909091e-06,
      "loss": 184.4307,
      "step": 10530
    },
    {
      "epoch": 0.08719030483517393,
      "grad_norm": 1035.8701171875,
      "learning_rate": 8.517171717171718e-06,
      "loss": 228.6084,
      "step": 10540
    },
    {
      "epoch": 0.08727302808454317,
      "grad_norm": 668.9830932617188,
      "learning_rate": 8.525252525252527e-06,
      "loss": 200.758,
      "step": 10550
    },
    {
      "epoch": 0.0873557513339124,
      "grad_norm": 1058.86181640625,
      "learning_rate": 8.533333333333335e-06,
      "loss": 204.7663,
      "step": 10560
    },
    {
      "epoch": 0.08743847458328163,
      "grad_norm": 1521.737548828125,
      "learning_rate": 8.541414141414142e-06,
      "loss": 192.2928,
      "step": 10570
    },
    {
      "epoch": 0.08752119783265086,
      "grad_norm": 1117.1220703125,
      "learning_rate": 8.54949494949495e-06,
      "loss": 142.4138,
      "step": 10580
    },
    {
      "epoch": 0.0876039210820201,
      "grad_norm": 1011.63525390625,
      "learning_rate": 8.557575757575757e-06,
      "loss": 193.8034,
      "step": 10590
    },
    {
      "epoch": 0.08768664433138934,
      "grad_norm": 2114.847900390625,
      "learning_rate": 8.565656565656566e-06,
      "loss": 184.3933,
      "step": 10600
    },
    {
      "epoch": 0.08776936758075857,
      "grad_norm": 823.4564819335938,
      "learning_rate": 8.573737373737374e-06,
      "loss": 186.983,
      "step": 10610
    },
    {
      "epoch": 0.08785209083012781,
      "grad_norm": 1262.8677978515625,
      "learning_rate": 8.581818181818183e-06,
      "loss": 145.9637,
      "step": 10620
    },
    {
      "epoch": 0.08793481407949705,
      "grad_norm": 1084.4609375,
      "learning_rate": 8.589898989898991e-06,
      "loss": 199.8638,
      "step": 10630
    },
    {
      "epoch": 0.08801753732886627,
      "grad_norm": 1483.7017822265625,
      "learning_rate": 8.597979797979798e-06,
      "loss": 183.0729,
      "step": 10640
    },
    {
      "epoch": 0.08810026057823551,
      "grad_norm": 1132.1685791015625,
      "learning_rate": 8.606060606060606e-06,
      "loss": 160.7905,
      "step": 10650
    },
    {
      "epoch": 0.08818298382760476,
      "grad_norm": 1912.802978515625,
      "learning_rate": 8.614141414141415e-06,
      "loss": 235.4324,
      "step": 10660
    },
    {
      "epoch": 0.08826570707697398,
      "grad_norm": 1315.4351806640625,
      "learning_rate": 8.622222222222223e-06,
      "loss": 186.3054,
      "step": 10670
    },
    {
      "epoch": 0.08834843032634322,
      "grad_norm": 884.4146118164062,
      "learning_rate": 8.630303030303032e-06,
      "loss": 157.2276,
      "step": 10680
    },
    {
      "epoch": 0.08843115357571245,
      "grad_norm": 1087.9329833984375,
      "learning_rate": 8.63838383838384e-06,
      "loss": 154.3768,
      "step": 10690
    },
    {
      "epoch": 0.08851387682508169,
      "grad_norm": 913.88818359375,
      "learning_rate": 8.646464646464647e-06,
      "loss": 185.6447,
      "step": 10700
    },
    {
      "epoch": 0.08859660007445093,
      "grad_norm": 2494.311279296875,
      "learning_rate": 8.654545454545455e-06,
      "loss": 195.7933,
      "step": 10710
    },
    {
      "epoch": 0.08867932332382016,
      "grad_norm": 1353.1029052734375,
      "learning_rate": 8.662626262626264e-06,
      "loss": 179.2149,
      "step": 10720
    },
    {
      "epoch": 0.0887620465731894,
      "grad_norm": 1235.08056640625,
      "learning_rate": 8.670707070707072e-06,
      "loss": 140.4072,
      "step": 10730
    },
    {
      "epoch": 0.08884476982255864,
      "grad_norm": 1513.452880859375,
      "learning_rate": 8.67878787878788e-06,
      "loss": 188.1772,
      "step": 10740
    },
    {
      "epoch": 0.08892749307192786,
      "grad_norm": 576.0892333984375,
      "learning_rate": 8.686868686868687e-06,
      "loss": 149.6942,
      "step": 10750
    },
    {
      "epoch": 0.0890102163212971,
      "grad_norm": 1103.5166015625,
      "learning_rate": 8.694949494949496e-06,
      "loss": 165.9619,
      "step": 10760
    },
    {
      "epoch": 0.08909293957066633,
      "grad_norm": 964.4625244140625,
      "learning_rate": 8.703030303030304e-06,
      "loss": 138.5367,
      "step": 10770
    },
    {
      "epoch": 0.08917566282003557,
      "grad_norm": 1614.7874755859375,
      "learning_rate": 8.711111111111111e-06,
      "loss": 175.2976,
      "step": 10780
    },
    {
      "epoch": 0.08925838606940481,
      "grad_norm": 1648.361328125,
      "learning_rate": 8.71919191919192e-06,
      "loss": 187.3021,
      "step": 10790
    },
    {
      "epoch": 0.08934110931877404,
      "grad_norm": 1707.2655029296875,
      "learning_rate": 8.727272727272728e-06,
      "loss": 258.8253,
      "step": 10800
    },
    {
      "epoch": 0.08942383256814328,
      "grad_norm": 1160.01806640625,
      "learning_rate": 8.735353535353536e-06,
      "loss": 194.5743,
      "step": 10810
    },
    {
      "epoch": 0.08950655581751252,
      "grad_norm": 1146.39990234375,
      "learning_rate": 8.743434343434343e-06,
      "loss": 203.5324,
      "step": 10820
    },
    {
      "epoch": 0.08958927906688174,
      "grad_norm": 3626.928955078125,
      "learning_rate": 8.751515151515151e-06,
      "loss": 253.2095,
      "step": 10830
    },
    {
      "epoch": 0.08967200231625098,
      "grad_norm": 728.5478515625,
      "learning_rate": 8.75959595959596e-06,
      "loss": 145.7956,
      "step": 10840
    },
    {
      "epoch": 0.08975472556562021,
      "grad_norm": 1002.9689331054688,
      "learning_rate": 8.767676767676768e-06,
      "loss": 181.9694,
      "step": 10850
    },
    {
      "epoch": 0.08983744881498945,
      "grad_norm": 1103.3865966796875,
      "learning_rate": 8.775757575757577e-06,
      "loss": 177.4934,
      "step": 10860
    },
    {
      "epoch": 0.08992017206435869,
      "grad_norm": 1987.7679443359375,
      "learning_rate": 8.783838383838385e-06,
      "loss": 190.5272,
      "step": 10870
    },
    {
      "epoch": 0.09000289531372792,
      "grad_norm": 1918.5572509765625,
      "learning_rate": 8.791919191919192e-06,
      "loss": 176.4213,
      "step": 10880
    },
    {
      "epoch": 0.09008561856309716,
      "grad_norm": 979.5933837890625,
      "learning_rate": 8.8e-06,
      "loss": 208.7932,
      "step": 10890
    },
    {
      "epoch": 0.0901683418124664,
      "grad_norm": 1254.18603515625,
      "learning_rate": 8.808080808080809e-06,
      "loss": 179.7351,
      "step": 10900
    },
    {
      "epoch": 0.09025106506183563,
      "grad_norm": 1393.46044921875,
      "learning_rate": 8.816161616161617e-06,
      "loss": 239.5334,
      "step": 10910
    },
    {
      "epoch": 0.09033378831120487,
      "grad_norm": 932.0426025390625,
      "learning_rate": 8.824242424242426e-06,
      "loss": 159.2398,
      "step": 10920
    },
    {
      "epoch": 0.0904165115605741,
      "grad_norm": 1175.749755859375,
      "learning_rate": 8.832323232323233e-06,
      "loss": 196.0824,
      "step": 10930
    },
    {
      "epoch": 0.09049923480994333,
      "grad_norm": 669.9056396484375,
      "learning_rate": 8.840404040404041e-06,
      "loss": 195.3154,
      "step": 10940
    },
    {
      "epoch": 0.09058195805931257,
      "grad_norm": 2161.71826171875,
      "learning_rate": 8.84848484848485e-06,
      "loss": 201.2208,
      "step": 10950
    },
    {
      "epoch": 0.0906646813086818,
      "grad_norm": 1083.76318359375,
      "learning_rate": 8.856565656565658e-06,
      "loss": 181.3205,
      "step": 10960
    },
    {
      "epoch": 0.09074740455805104,
      "grad_norm": 1360.2738037109375,
      "learning_rate": 8.864646464646466e-06,
      "loss": 173.5026,
      "step": 10970
    },
    {
      "epoch": 0.09083012780742028,
      "grad_norm": 1274.3690185546875,
      "learning_rate": 8.872727272727275e-06,
      "loss": 191.3013,
      "step": 10980
    },
    {
      "epoch": 0.0909128510567895,
      "grad_norm": 1275.59814453125,
      "learning_rate": 8.880808080808082e-06,
      "loss": 180.379,
      "step": 10990
    },
    {
      "epoch": 0.09099557430615875,
      "grad_norm": 1908.28125,
      "learning_rate": 8.888888888888888e-06,
      "loss": 226.928,
      "step": 11000
    },
    {
      "epoch": 0.09107829755552799,
      "grad_norm": 1636.098388671875,
      "learning_rate": 8.896969696969697e-06,
      "loss": 166.5137,
      "step": 11010
    },
    {
      "epoch": 0.09116102080489721,
      "grad_norm": 1362.2508544921875,
      "learning_rate": 8.905050505050505e-06,
      "loss": 162.6263,
      "step": 11020
    },
    {
      "epoch": 0.09124374405426645,
      "grad_norm": 958.8033447265625,
      "learning_rate": 8.913131313131314e-06,
      "loss": 126.4212,
      "step": 11030
    },
    {
      "epoch": 0.09132646730363568,
      "grad_norm": 4022.8896484375,
      "learning_rate": 8.921212121212122e-06,
      "loss": 171.8223,
      "step": 11040
    },
    {
      "epoch": 0.09140919055300492,
      "grad_norm": 1437.0355224609375,
      "learning_rate": 8.92929292929293e-06,
      "loss": 150.3858,
      "step": 11050
    },
    {
      "epoch": 0.09149191380237416,
      "grad_norm": 1457.9029541015625,
      "learning_rate": 8.937373737373737e-06,
      "loss": 191.1608,
      "step": 11060
    },
    {
      "epoch": 0.09157463705174339,
      "grad_norm": 1079.6673583984375,
      "learning_rate": 8.945454545454546e-06,
      "loss": 169.316,
      "step": 11070
    },
    {
      "epoch": 0.09165736030111263,
      "grad_norm": 1028.742431640625,
      "learning_rate": 8.953535353535354e-06,
      "loss": 141.8136,
      "step": 11080
    },
    {
      "epoch": 0.09174008355048187,
      "grad_norm": 1539.916259765625,
      "learning_rate": 8.961616161616163e-06,
      "loss": 196.1172,
      "step": 11090
    },
    {
      "epoch": 0.0918228067998511,
      "grad_norm": 1310.006591796875,
      "learning_rate": 8.969696969696971e-06,
      "loss": 153.3139,
      "step": 11100
    },
    {
      "epoch": 0.09190553004922034,
      "grad_norm": 3632.826416015625,
      "learning_rate": 8.977777777777778e-06,
      "loss": 274.3921,
      "step": 11110
    },
    {
      "epoch": 0.09198825329858956,
      "grad_norm": 1360.5455322265625,
      "learning_rate": 8.985858585858586e-06,
      "loss": 199.4904,
      "step": 11120
    },
    {
      "epoch": 0.0920709765479588,
      "grad_norm": 1219.487548828125,
      "learning_rate": 8.993939393939395e-06,
      "loss": 138.8559,
      "step": 11130
    },
    {
      "epoch": 0.09215369979732804,
      "grad_norm": 1320.4920654296875,
      "learning_rate": 9.002020202020203e-06,
      "loss": 179.0958,
      "step": 11140
    },
    {
      "epoch": 0.09223642304669727,
      "grad_norm": 754.0531616210938,
      "learning_rate": 9.010101010101012e-06,
      "loss": 183.8919,
      "step": 11150
    },
    {
      "epoch": 0.09231914629606651,
      "grad_norm": 1132.8018798828125,
      "learning_rate": 9.01818181818182e-06,
      "loss": 138.0276,
      "step": 11160
    },
    {
      "epoch": 0.09240186954543575,
      "grad_norm": 1015.0302124023438,
      "learning_rate": 9.026262626262627e-06,
      "loss": 169.833,
      "step": 11170
    },
    {
      "epoch": 0.09248459279480498,
      "grad_norm": 1256.841552734375,
      "learning_rate": 9.034343434343435e-06,
      "loss": 192.5117,
      "step": 11180
    },
    {
      "epoch": 0.09256731604417422,
      "grad_norm": 637.6029663085938,
      "learning_rate": 9.042424242424244e-06,
      "loss": 136.4471,
      "step": 11190
    },
    {
      "epoch": 0.09265003929354346,
      "grad_norm": 2063.305419921875,
      "learning_rate": 9.050505050505052e-06,
      "loss": 187.3035,
      "step": 11200
    },
    {
      "epoch": 0.09273276254291268,
      "grad_norm": 893.4026489257812,
      "learning_rate": 9.058585858585859e-06,
      "loss": 197.6645,
      "step": 11210
    },
    {
      "epoch": 0.09281548579228192,
      "grad_norm": 11727.92578125,
      "learning_rate": 9.066666666666667e-06,
      "loss": 184.4137,
      "step": 11220
    },
    {
      "epoch": 0.09289820904165115,
      "grad_norm": 939.4194946289062,
      "learning_rate": 9.074747474747476e-06,
      "loss": 163.7697,
      "step": 11230
    },
    {
      "epoch": 0.09298093229102039,
      "grad_norm": 1211.7532958984375,
      "learning_rate": 9.082828282828283e-06,
      "loss": 150.1024,
      "step": 11240
    },
    {
      "epoch": 0.09306365554038963,
      "grad_norm": 1020.1021118164062,
      "learning_rate": 9.090909090909091e-06,
      "loss": 176.2725,
      "step": 11250
    },
    {
      "epoch": 0.09314637878975886,
      "grad_norm": 1597.582763671875,
      "learning_rate": 9.0989898989899e-06,
      "loss": 161.5863,
      "step": 11260
    },
    {
      "epoch": 0.0932291020391281,
      "grad_norm": 1205.4434814453125,
      "learning_rate": 9.107070707070708e-06,
      "loss": 191.3436,
      "step": 11270
    },
    {
      "epoch": 0.09331182528849734,
      "grad_norm": 1347.427001953125,
      "learning_rate": 9.115151515151516e-06,
      "loss": 147.056,
      "step": 11280
    },
    {
      "epoch": 0.09339454853786656,
      "grad_norm": 1340.0323486328125,
      "learning_rate": 9.123232323232323e-06,
      "loss": 166.0659,
      "step": 11290
    },
    {
      "epoch": 0.0934772717872358,
      "grad_norm": 1243.469482421875,
      "learning_rate": 9.131313131313132e-06,
      "loss": 160.2771,
      "step": 11300
    },
    {
      "epoch": 0.09355999503660503,
      "grad_norm": 1073.79638671875,
      "learning_rate": 9.13939393939394e-06,
      "loss": 189.4957,
      "step": 11310
    },
    {
      "epoch": 0.09364271828597427,
      "grad_norm": 1185.810546875,
      "learning_rate": 9.147474747474748e-06,
      "loss": 169.7764,
      "step": 11320
    },
    {
      "epoch": 0.09372544153534351,
      "grad_norm": 1002.9879760742188,
      "learning_rate": 9.155555555555557e-06,
      "loss": 117.5662,
      "step": 11330
    },
    {
      "epoch": 0.09380816478471274,
      "grad_norm": 1063.5313720703125,
      "learning_rate": 9.163636363636365e-06,
      "loss": 199.0167,
      "step": 11340
    },
    {
      "epoch": 0.09389088803408198,
      "grad_norm": 975.7847290039062,
      "learning_rate": 9.171717171717172e-06,
      "loss": 178.3596,
      "step": 11350
    },
    {
      "epoch": 0.09397361128345122,
      "grad_norm": 1078.6475830078125,
      "learning_rate": 9.17979797979798e-06,
      "loss": 167.8156,
      "step": 11360
    },
    {
      "epoch": 0.09405633453282045,
      "grad_norm": 2150.571044921875,
      "learning_rate": 9.187878787878789e-06,
      "loss": 151.6369,
      "step": 11370
    },
    {
      "epoch": 0.09413905778218969,
      "grad_norm": 1107.6488037109375,
      "learning_rate": 9.195959595959597e-06,
      "loss": 138.8753,
      "step": 11380
    },
    {
      "epoch": 0.09422178103155891,
      "grad_norm": 2013.7427978515625,
      "learning_rate": 9.204040404040406e-06,
      "loss": 168.8607,
      "step": 11390
    },
    {
      "epoch": 0.09430450428092815,
      "grad_norm": 926.5513916015625,
      "learning_rate": 9.212121212121213e-06,
      "loss": 155.2859,
      "step": 11400
    },
    {
      "epoch": 0.0943872275302974,
      "grad_norm": 640.766357421875,
      "learning_rate": 9.220202020202021e-06,
      "loss": 170.4931,
      "step": 11410
    },
    {
      "epoch": 0.09446995077966662,
      "grad_norm": 2521.883544921875,
      "learning_rate": 9.228282828282828e-06,
      "loss": 163.8456,
      "step": 11420
    },
    {
      "epoch": 0.09455267402903586,
      "grad_norm": 2574.5419921875,
      "learning_rate": 9.236363636363636e-06,
      "loss": 192.8077,
      "step": 11430
    },
    {
      "epoch": 0.0946353972784051,
      "grad_norm": 1514.116943359375,
      "learning_rate": 9.244444444444445e-06,
      "loss": 190.4653,
      "step": 11440
    },
    {
      "epoch": 0.09471812052777433,
      "grad_norm": 1116.5655517578125,
      "learning_rate": 9.252525252525253e-06,
      "loss": 150.5008,
      "step": 11450
    },
    {
      "epoch": 0.09480084377714357,
      "grad_norm": 7808.6015625,
      "learning_rate": 9.260606060606062e-06,
      "loss": 282.1488,
      "step": 11460
    },
    {
      "epoch": 0.09488356702651281,
      "grad_norm": 1106.9825439453125,
      "learning_rate": 9.268686868686868e-06,
      "loss": 195.7642,
      "step": 11470
    },
    {
      "epoch": 0.09496629027588203,
      "grad_norm": 3581.72900390625,
      "learning_rate": 9.276767676767677e-06,
      "loss": 179.7523,
      "step": 11480
    },
    {
      "epoch": 0.09504901352525127,
      "grad_norm": 1101.411865234375,
      "learning_rate": 9.284848484848485e-06,
      "loss": 199.7394,
      "step": 11490
    },
    {
      "epoch": 0.0951317367746205,
      "grad_norm": 1929.9052734375,
      "learning_rate": 9.292929292929294e-06,
      "loss": 174.8564,
      "step": 11500
    },
    {
      "epoch": 0.09521446002398974,
      "grad_norm": 1121.5028076171875,
      "learning_rate": 9.301010101010102e-06,
      "loss": 134.3296,
      "step": 11510
    },
    {
      "epoch": 0.09529718327335898,
      "grad_norm": 1080.38671875,
      "learning_rate": 9.30909090909091e-06,
      "loss": 173.9918,
      "step": 11520
    },
    {
      "epoch": 0.09537990652272821,
      "grad_norm": 1371.1961669921875,
      "learning_rate": 9.317171717171717e-06,
      "loss": 213.5175,
      "step": 11530
    },
    {
      "epoch": 0.09546262977209745,
      "grad_norm": 1480.9495849609375,
      "learning_rate": 9.325252525252526e-06,
      "loss": 136.4908,
      "step": 11540
    },
    {
      "epoch": 0.09554535302146669,
      "grad_norm": 912.1797485351562,
      "learning_rate": 9.333333333333334e-06,
      "loss": 215.026,
      "step": 11550
    },
    {
      "epoch": 0.09562807627083592,
      "grad_norm": 2468.00341796875,
      "learning_rate": 9.341414141414143e-06,
      "loss": 175.5986,
      "step": 11560
    },
    {
      "epoch": 0.09571079952020516,
      "grad_norm": 1296.7786865234375,
      "learning_rate": 9.349494949494951e-06,
      "loss": 209.1408,
      "step": 11570
    },
    {
      "epoch": 0.09579352276957438,
      "grad_norm": 1264.5654296875,
      "learning_rate": 9.357575757575758e-06,
      "loss": 200.8274,
      "step": 11580
    },
    {
      "epoch": 0.09587624601894362,
      "grad_norm": 1155.8309326171875,
      "learning_rate": 9.365656565656566e-06,
      "loss": 164.7913,
      "step": 11590
    },
    {
      "epoch": 0.09595896926831286,
      "grad_norm": 1020.3081665039062,
      "learning_rate": 9.373737373737375e-06,
      "loss": 155.3867,
      "step": 11600
    },
    {
      "epoch": 0.09604169251768209,
      "grad_norm": 1648.6768798828125,
      "learning_rate": 9.381818181818183e-06,
      "loss": 150.0429,
      "step": 11610
    },
    {
      "epoch": 0.09612441576705133,
      "grad_norm": 1153.5174560546875,
      "learning_rate": 9.389898989898992e-06,
      "loss": 183.8605,
      "step": 11620
    },
    {
      "epoch": 0.09620713901642057,
      "grad_norm": 1894.8492431640625,
      "learning_rate": 9.397979797979799e-06,
      "loss": 178.0279,
      "step": 11630
    },
    {
      "epoch": 0.0962898622657898,
      "grad_norm": 1127.750244140625,
      "learning_rate": 9.406060606060607e-06,
      "loss": 162.6299,
      "step": 11640
    },
    {
      "epoch": 0.09637258551515904,
      "grad_norm": 1268.74267578125,
      "learning_rate": 9.414141414141414e-06,
      "loss": 155.4865,
      "step": 11650
    },
    {
      "epoch": 0.09645530876452828,
      "grad_norm": 1007.6961669921875,
      "learning_rate": 9.422222222222222e-06,
      "loss": 209.233,
      "step": 11660
    },
    {
      "epoch": 0.0965380320138975,
      "grad_norm": 1457.4285888671875,
      "learning_rate": 9.43030303030303e-06,
      "loss": 163.3144,
      "step": 11670
    },
    {
      "epoch": 0.09662075526326674,
      "grad_norm": 1042.603515625,
      "learning_rate": 9.438383838383839e-06,
      "loss": 182.3341,
      "step": 11680
    },
    {
      "epoch": 0.09670347851263597,
      "grad_norm": 1047.1702880859375,
      "learning_rate": 9.446464646464648e-06,
      "loss": 167.2329,
      "step": 11690
    },
    {
      "epoch": 0.09678620176200521,
      "grad_norm": 1840.010009765625,
      "learning_rate": 9.454545454545456e-06,
      "loss": 204.4755,
      "step": 11700
    },
    {
      "epoch": 0.09686892501137445,
      "grad_norm": 573.9185180664062,
      "learning_rate": 9.462626262626263e-06,
      "loss": 106.0682,
      "step": 11710
    },
    {
      "epoch": 0.09695164826074368,
      "grad_norm": 1605.5093994140625,
      "learning_rate": 9.470707070707071e-06,
      "loss": 214.7734,
      "step": 11720
    },
    {
      "epoch": 0.09703437151011292,
      "grad_norm": 2460.1376953125,
      "learning_rate": 9.47878787878788e-06,
      "loss": 180.7382,
      "step": 11730
    },
    {
      "epoch": 0.09711709475948216,
      "grad_norm": 1100.57080078125,
      "learning_rate": 9.486868686868688e-06,
      "loss": 220.1118,
      "step": 11740
    },
    {
      "epoch": 0.09719981800885139,
      "grad_norm": 937.7584838867188,
      "learning_rate": 9.494949494949497e-06,
      "loss": 176.8332,
      "step": 11750
    },
    {
      "epoch": 0.09728254125822063,
      "grad_norm": 1534.452880859375,
      "learning_rate": 9.503030303030303e-06,
      "loss": 173.9935,
      "step": 11760
    },
    {
      "epoch": 0.09736526450758985,
      "grad_norm": 1466.6376953125,
      "learning_rate": 9.511111111111112e-06,
      "loss": 196.6087,
      "step": 11770
    },
    {
      "epoch": 0.09744798775695909,
      "grad_norm": 1731.87841796875,
      "learning_rate": 9.51919191919192e-06,
      "loss": 179.0696,
      "step": 11780
    },
    {
      "epoch": 0.09753071100632833,
      "grad_norm": 1235.0560302734375,
      "learning_rate": 9.527272727272729e-06,
      "loss": 164.9756,
      "step": 11790
    },
    {
      "epoch": 0.09761343425569756,
      "grad_norm": 662.07568359375,
      "learning_rate": 9.535353535353537e-06,
      "loss": 155.2859,
      "step": 11800
    },
    {
      "epoch": 0.0976961575050668,
      "grad_norm": 642.6212158203125,
      "learning_rate": 9.543434343434344e-06,
      "loss": 183.2646,
      "step": 11810
    },
    {
      "epoch": 0.09777888075443604,
      "grad_norm": 861.9931030273438,
      "learning_rate": 9.551515151515152e-06,
      "loss": 140.6058,
      "step": 11820
    },
    {
      "epoch": 0.09786160400380527,
      "grad_norm": 2064.848388671875,
      "learning_rate": 9.55959595959596e-06,
      "loss": 175.8456,
      "step": 11830
    },
    {
      "epoch": 0.0979443272531745,
      "grad_norm": 1402.7391357421875,
      "learning_rate": 9.56767676767677e-06,
      "loss": 168.1851,
      "step": 11840
    },
    {
      "epoch": 0.09802705050254373,
      "grad_norm": 1274.8916015625,
      "learning_rate": 9.575757575757576e-06,
      "loss": 177.3613,
      "step": 11850
    },
    {
      "epoch": 0.09810977375191297,
      "grad_norm": 1175.4793701171875,
      "learning_rate": 9.583838383838384e-06,
      "loss": 155.9946,
      "step": 11860
    },
    {
      "epoch": 0.09819249700128221,
      "grad_norm": 1113.3656005859375,
      "learning_rate": 9.591919191919193e-06,
      "loss": 174.7684,
      "step": 11870
    },
    {
      "epoch": 0.09827522025065144,
      "grad_norm": 1227.9544677734375,
      "learning_rate": 9.600000000000001e-06,
      "loss": 173.7199,
      "step": 11880
    },
    {
      "epoch": 0.09835794350002068,
      "grad_norm": 1131.5029296875,
      "learning_rate": 9.608080808080808e-06,
      "loss": 179.6841,
      "step": 11890
    },
    {
      "epoch": 0.09844066674938992,
      "grad_norm": 1879.9422607421875,
      "learning_rate": 9.616161616161616e-06,
      "loss": 201.8642,
      "step": 11900
    },
    {
      "epoch": 0.09852338999875915,
      "grad_norm": 1383.9271240234375,
      "learning_rate": 9.624242424242425e-06,
      "loss": 155.471,
      "step": 11910
    },
    {
      "epoch": 0.09860611324812839,
      "grad_norm": 1202.215576171875,
      "learning_rate": 9.632323232323233e-06,
      "loss": 176.2691,
      "step": 11920
    },
    {
      "epoch": 0.09868883649749763,
      "grad_norm": 713.47216796875,
      "learning_rate": 9.640404040404042e-06,
      "loss": 129.0435,
      "step": 11930
    },
    {
      "epoch": 0.09877155974686685,
      "grad_norm": 1676.49658203125,
      "learning_rate": 9.648484848484849e-06,
      "loss": 158.0141,
      "step": 11940
    },
    {
      "epoch": 0.0988542829962361,
      "grad_norm": 822.8709716796875,
      "learning_rate": 9.656565656565657e-06,
      "loss": 128.7683,
      "step": 11950
    },
    {
      "epoch": 0.09893700624560532,
      "grad_norm": 2505.596923828125,
      "learning_rate": 9.664646464646465e-06,
      "loss": 173.4101,
      "step": 11960
    },
    {
      "epoch": 0.09901972949497456,
      "grad_norm": 764.0625610351562,
      "learning_rate": 9.672727272727274e-06,
      "loss": 186.7536,
      "step": 11970
    },
    {
      "epoch": 0.0991024527443438,
      "grad_norm": 1434.5181884765625,
      "learning_rate": 9.680808080808082e-06,
      "loss": 195.2917,
      "step": 11980
    },
    {
      "epoch": 0.09918517599371303,
      "grad_norm": 916.0157470703125,
      "learning_rate": 9.688888888888889e-06,
      "loss": 151.0252,
      "step": 11990
    },
    {
      "epoch": 0.09926789924308227,
      "grad_norm": 1034.74072265625,
      "learning_rate": 9.696969696969698e-06,
      "loss": 163.2282,
      "step": 12000
    },
    {
      "epoch": 0.09935062249245151,
      "grad_norm": 1380.8863525390625,
      "learning_rate": 9.705050505050506e-06,
      "loss": 139.9053,
      "step": 12010
    },
    {
      "epoch": 0.09943334574182074,
      "grad_norm": 911.8162231445312,
      "learning_rate": 9.713131313131314e-06,
      "loss": 159.4434,
      "step": 12020
    },
    {
      "epoch": 0.09951606899118998,
      "grad_norm": 1710.5338134765625,
      "learning_rate": 9.721212121212123e-06,
      "loss": 152.8905,
      "step": 12030
    },
    {
      "epoch": 0.0995987922405592,
      "grad_norm": 1827.9671630859375,
      "learning_rate": 9.729292929292931e-06,
      "loss": 195.7561,
      "step": 12040
    },
    {
      "epoch": 0.09968151548992844,
      "grad_norm": 1594.53759765625,
      "learning_rate": 9.737373737373738e-06,
      "loss": 172.4959,
      "step": 12050
    },
    {
      "epoch": 0.09976423873929768,
      "grad_norm": 985.9189453125,
      "learning_rate": 9.745454545454547e-06,
      "loss": 167.3074,
      "step": 12060
    },
    {
      "epoch": 0.09984696198866691,
      "grad_norm": 1075.2579345703125,
      "learning_rate": 9.753535353535353e-06,
      "loss": 163.0599,
      "step": 12070
    },
    {
      "epoch": 0.09992968523803615,
      "grad_norm": 2030.679443359375,
      "learning_rate": 9.761616161616162e-06,
      "loss": 163.872,
      "step": 12080
    },
    {
      "epoch": 0.10001240848740539,
      "grad_norm": 1216.58984375,
      "learning_rate": 9.76969696969697e-06,
      "loss": 158.4474,
      "step": 12090
    },
    {
      "epoch": 0.10009513173677462,
      "grad_norm": 1842.97998046875,
      "learning_rate": 9.777777777777779e-06,
      "loss": 225.2362,
      "step": 12100
    },
    {
      "epoch": 0.10017785498614386,
      "grad_norm": 1359.1461181640625,
      "learning_rate": 9.785858585858587e-06,
      "loss": 172.9057,
      "step": 12110
    },
    {
      "epoch": 0.10026057823551308,
      "grad_norm": 1055.8148193359375,
      "learning_rate": 9.793939393939394e-06,
      "loss": 153.7417,
      "step": 12120
    },
    {
      "epoch": 0.10034330148488232,
      "grad_norm": 1045.6163330078125,
      "learning_rate": 9.802020202020202e-06,
      "loss": 151.8674,
      "step": 12130
    },
    {
      "epoch": 0.10042602473425156,
      "grad_norm": 702.3483276367188,
      "learning_rate": 9.81010101010101e-06,
      "loss": 154.9376,
      "step": 12140
    },
    {
      "epoch": 0.10050874798362079,
      "grad_norm": 582.2290649414062,
      "learning_rate": 9.81818181818182e-06,
      "loss": 191.1368,
      "step": 12150
    },
    {
      "epoch": 0.10059147123299003,
      "grad_norm": 1067.549072265625,
      "learning_rate": 9.826262626262628e-06,
      "loss": 215.4936,
      "step": 12160
    },
    {
      "epoch": 0.10067419448235927,
      "grad_norm": 1477.08251953125,
      "learning_rate": 9.834343434343434e-06,
      "loss": 179.9154,
      "step": 12170
    },
    {
      "epoch": 0.1007569177317285,
      "grad_norm": 1615.7445068359375,
      "learning_rate": 9.842424242424243e-06,
      "loss": 151.6433,
      "step": 12180
    },
    {
      "epoch": 0.10083964098109774,
      "grad_norm": 1283.7108154296875,
      "learning_rate": 9.850505050505051e-06,
      "loss": 176.6817,
      "step": 12190
    },
    {
      "epoch": 0.10092236423046698,
      "grad_norm": 1741.5172119140625,
      "learning_rate": 9.85858585858586e-06,
      "loss": 148.4978,
      "step": 12200
    },
    {
      "epoch": 0.1010050874798362,
      "grad_norm": 956.2932739257812,
      "learning_rate": 9.866666666666668e-06,
      "loss": 178.0466,
      "step": 12210
    },
    {
      "epoch": 0.10108781072920545,
      "grad_norm": 771.7984008789062,
      "learning_rate": 9.874747474747477e-06,
      "loss": 177.3261,
      "step": 12220
    },
    {
      "epoch": 0.10117053397857467,
      "grad_norm": 703.4427490234375,
      "learning_rate": 9.882828282828283e-06,
      "loss": 175.3549,
      "step": 12230
    },
    {
      "epoch": 0.10125325722794391,
      "grad_norm": 931.7167358398438,
      "learning_rate": 9.890909090909092e-06,
      "loss": 158.6873,
      "step": 12240
    },
    {
      "epoch": 0.10133598047731315,
      "grad_norm": 1080.5570068359375,
      "learning_rate": 9.8989898989899e-06,
      "loss": 109.4355,
      "step": 12250
    },
    {
      "epoch": 0.10141870372668238,
      "grad_norm": 1163.09423828125,
      "learning_rate": 9.907070707070709e-06,
      "loss": 164.6523,
      "step": 12260
    },
    {
      "epoch": 0.10150142697605162,
      "grad_norm": 870.4569091796875,
      "learning_rate": 9.915151515151515e-06,
      "loss": 162.8174,
      "step": 12270
    },
    {
      "epoch": 0.10158415022542086,
      "grad_norm": 976.0827026367188,
      "learning_rate": 9.923232323232324e-06,
      "loss": 149.6927,
      "step": 12280
    },
    {
      "epoch": 0.10166687347479009,
      "grad_norm": 881.8626098632812,
      "learning_rate": 9.931313131313132e-06,
      "loss": 128.7362,
      "step": 12290
    },
    {
      "epoch": 0.10174959672415933,
      "grad_norm": 612.5805053710938,
      "learning_rate": 9.939393939393939e-06,
      "loss": 129.0437,
      "step": 12300
    },
    {
      "epoch": 0.10183231997352855,
      "grad_norm": 1372.064453125,
      "learning_rate": 9.947474747474748e-06,
      "loss": 188.9469,
      "step": 12310
    },
    {
      "epoch": 0.1019150432228978,
      "grad_norm": 1295.59375,
      "learning_rate": 9.955555555555556e-06,
      "loss": 167.0848,
      "step": 12320
    },
    {
      "epoch": 0.10199776647226703,
      "grad_norm": 1037.228515625,
      "learning_rate": 9.963636363636364e-06,
      "loss": 176.8787,
      "step": 12330
    },
    {
      "epoch": 0.10208048972163626,
      "grad_norm": 1227.047607421875,
      "learning_rate": 9.971717171717173e-06,
      "loss": 198.2504,
      "step": 12340
    },
    {
      "epoch": 0.1021632129710055,
      "grad_norm": 1416.220458984375,
      "learning_rate": 9.97979797979798e-06,
      "loss": 179.9958,
      "step": 12350
    },
    {
      "epoch": 0.10224593622037474,
      "grad_norm": 1424.654052734375,
      "learning_rate": 9.987878787878788e-06,
      "loss": 191.8388,
      "step": 12360
    },
    {
      "epoch": 0.10232865946974397,
      "grad_norm": 1317.3236083984375,
      "learning_rate": 9.995959595959597e-06,
      "loss": 167.6917,
      "step": 12370
    },
    {
      "epoch": 0.10241138271911321,
      "grad_norm": 868.8121337890625,
      "learning_rate": 9.99999995027162e-06,
      "loss": 156.1878,
      "step": 12380
    },
    {
      "epoch": 0.10249410596848244,
      "grad_norm": 889.690185546875,
      "learning_rate": 9.99999955244457e-06,
      "loss": 188.0309,
      "step": 12390
    },
    {
      "epoch": 0.10257682921785168,
      "grad_norm": 964.9615478515625,
      "learning_rate": 9.999998756790503e-06,
      "loss": 162.0736,
      "step": 12400
    },
    {
      "epoch": 0.10265955246722092,
      "grad_norm": 2292.847900390625,
      "learning_rate": 9.999997563309483e-06,
      "loss": 138.4502,
      "step": 12410
    },
    {
      "epoch": 0.10274227571659014,
      "grad_norm": 1073.1864013671875,
      "learning_rate": 9.999995972001602e-06,
      "loss": 183.7101,
      "step": 12420
    },
    {
      "epoch": 0.10282499896595938,
      "grad_norm": 1411.09521484375,
      "learning_rate": 9.99999398286699e-06,
      "loss": 172.5208,
      "step": 12430
    },
    {
      "epoch": 0.10290772221532862,
      "grad_norm": 1048.4619140625,
      "learning_rate": 9.999991595905803e-06,
      "loss": 169.999,
      "step": 12440
    },
    {
      "epoch": 0.10299044546469785,
      "grad_norm": 986.3778686523438,
      "learning_rate": 9.999988811118232e-06,
      "loss": 146.7089,
      "step": 12450
    },
    {
      "epoch": 0.10307316871406709,
      "grad_norm": 795.2116088867188,
      "learning_rate": 9.999985628504498e-06,
      "loss": 159.4188,
      "step": 12460
    },
    {
      "epoch": 0.10315589196343633,
      "grad_norm": 919.0172119140625,
      "learning_rate": 9.999982048064854e-06,
      "loss": 144.6598,
      "step": 12470
    },
    {
      "epoch": 0.10323861521280556,
      "grad_norm": 1297.25732421875,
      "learning_rate": 9.999978069799585e-06,
      "loss": 171.9091,
      "step": 12480
    },
    {
      "epoch": 0.1033213384621748,
      "grad_norm": 1352.2923583984375,
      "learning_rate": 9.999973693709008e-06,
      "loss": 151.0232,
      "step": 12490
    },
    {
      "epoch": 0.10340406171154402,
      "grad_norm": 1723.31640625,
      "learning_rate": 9.99996891979347e-06,
      "loss": 131.0583,
      "step": 12500
    },
    {
      "epoch": 0.10348678496091326,
      "grad_norm": 1077.4090576171875,
      "learning_rate": 9.999963748053354e-06,
      "loss": 155.0065,
      "step": 12510
    },
    {
      "epoch": 0.1035695082102825,
      "grad_norm": 881.5154418945312,
      "learning_rate": 9.999958178489069e-06,
      "loss": 150.2538,
      "step": 12520
    },
    {
      "epoch": 0.10365223145965173,
      "grad_norm": 1021.1143188476562,
      "learning_rate": 9.999952211101056e-06,
      "loss": 154.2661,
      "step": 12530
    },
    {
      "epoch": 0.10373495470902097,
      "grad_norm": 1220.9776611328125,
      "learning_rate": 9.999945845889795e-06,
      "loss": 154.2683,
      "step": 12540
    },
    {
      "epoch": 0.10381767795839021,
      "grad_norm": 1148.5760498046875,
      "learning_rate": 9.999939082855788e-06,
      "loss": 137.0497,
      "step": 12550
    },
    {
      "epoch": 0.10390040120775944,
      "grad_norm": 1300.0975341796875,
      "learning_rate": 9.999931921999575e-06,
      "loss": 169.4536,
      "step": 12560
    },
    {
      "epoch": 0.10398312445712868,
      "grad_norm": 1183.8826904296875,
      "learning_rate": 9.999924363321726e-06,
      "loss": 166.6497,
      "step": 12570
    },
    {
      "epoch": 0.1040658477064979,
      "grad_norm": 718.7395629882812,
      "learning_rate": 9.999916406822843e-06,
      "loss": 109.411,
      "step": 12580
    },
    {
      "epoch": 0.10414857095586715,
      "grad_norm": 841.6725463867188,
      "learning_rate": 9.999908052503557e-06,
      "loss": 146.8344,
      "step": 12590
    },
    {
      "epoch": 0.10423129420523639,
      "grad_norm": 918.6251831054688,
      "learning_rate": 9.999899300364534e-06,
      "loss": 162.9566,
      "step": 12600
    },
    {
      "epoch": 0.10431401745460561,
      "grad_norm": 1467.325927734375,
      "learning_rate": 9.99989015040647e-06,
      "loss": 189.8623,
      "step": 12610
    },
    {
      "epoch": 0.10439674070397485,
      "grad_norm": 2947.025146484375,
      "learning_rate": 9.999880602630092e-06,
      "loss": 141.2983,
      "step": 12620
    },
    {
      "epoch": 0.10447946395334409,
      "grad_norm": 1483.4896240234375,
      "learning_rate": 9.999870657036161e-06,
      "loss": 162.2035,
      "step": 12630
    },
    {
      "epoch": 0.10456218720271332,
      "grad_norm": 1846.94580078125,
      "learning_rate": 9.99986031362547e-06,
      "loss": 150.0733,
      "step": 12640
    },
    {
      "epoch": 0.10464491045208256,
      "grad_norm": 1538.964111328125,
      "learning_rate": 9.99984957239884e-06,
      "loss": 206.5021,
      "step": 12650
    },
    {
      "epoch": 0.1047276337014518,
      "grad_norm": 1100.305908203125,
      "learning_rate": 9.999838433357124e-06,
      "loss": 197.57,
      "step": 12660
    },
    {
      "epoch": 0.10481035695082103,
      "grad_norm": 594.6061401367188,
      "learning_rate": 9.99982689650121e-06,
      "loss": 179.5414,
      "step": 12670
    },
    {
      "epoch": 0.10489308020019027,
      "grad_norm": 737.6015014648438,
      "learning_rate": 9.999814961832018e-06,
      "loss": 170.4644,
      "step": 12680
    },
    {
      "epoch": 0.1049758034495595,
      "grad_norm": 1346.4503173828125,
      "learning_rate": 9.999802629350492e-06,
      "loss": 202.0369,
      "step": 12690
    },
    {
      "epoch": 0.10505852669892873,
      "grad_norm": 2559.85546875,
      "learning_rate": 9.99978989905762e-06,
      "loss": 221.5045,
      "step": 12700
    },
    {
      "epoch": 0.10514124994829797,
      "grad_norm": 1314.4476318359375,
      "learning_rate": 9.999776770954411e-06,
      "loss": 173.8742,
      "step": 12710
    },
    {
      "epoch": 0.1052239731976672,
      "grad_norm": 3551.19677734375,
      "learning_rate": 9.99976324504191e-06,
      "loss": 149.5013,
      "step": 12720
    },
    {
      "epoch": 0.10530669644703644,
      "grad_norm": 1428.34814453125,
      "learning_rate": 9.999749321321192e-06,
      "loss": 251.0179,
      "step": 12730
    },
    {
      "epoch": 0.10538941969640568,
      "grad_norm": 753.5355224609375,
      "learning_rate": 9.999734999793369e-06,
      "loss": 163.6853,
      "step": 12740
    },
    {
      "epoch": 0.10547214294577491,
      "grad_norm": 1578.79541015625,
      "learning_rate": 9.999720280459576e-06,
      "loss": 154.6416,
      "step": 12750
    },
    {
      "epoch": 0.10555486619514415,
      "grad_norm": 1111.734130859375,
      "learning_rate": 9.999705163320987e-06,
      "loss": 177.3941,
      "step": 12760
    },
    {
      "epoch": 0.10563758944451337,
      "grad_norm": 1722.9654541015625,
      "learning_rate": 9.999689648378801e-06,
      "loss": 179.8888,
      "step": 12770
    },
    {
      "epoch": 0.10572031269388261,
      "grad_norm": 1038.706298828125,
      "learning_rate": 9.999673735634259e-06,
      "loss": 127.1906,
      "step": 12780
    },
    {
      "epoch": 0.10580303594325186,
      "grad_norm": 11962.8779296875,
      "learning_rate": 9.99965742508862e-06,
      "loss": 213.939,
      "step": 12790
    },
    {
      "epoch": 0.10588575919262108,
      "grad_norm": 1482.868408203125,
      "learning_rate": 9.999640716743186e-06,
      "loss": 133.2547,
      "step": 12800
    },
    {
      "epoch": 0.10596848244199032,
      "grad_norm": 3282.25390625,
      "learning_rate": 9.999623610599287e-06,
      "loss": 160.013,
      "step": 12810
    },
    {
      "epoch": 0.10605120569135956,
      "grad_norm": 1836.9112548828125,
      "learning_rate": 9.999606106658282e-06,
      "loss": 201.4363,
      "step": 12820
    },
    {
      "epoch": 0.10613392894072879,
      "grad_norm": 816.5084838867188,
      "learning_rate": 9.999588204921562e-06,
      "loss": 174.8686,
      "step": 12830
    },
    {
      "epoch": 0.10621665219009803,
      "grad_norm": 1352.0396728515625,
      "learning_rate": 9.999569905390556e-06,
      "loss": 167.4276,
      "step": 12840
    },
    {
      "epoch": 0.10629937543946726,
      "grad_norm": 533.4453125,
      "learning_rate": 9.999551208066716e-06,
      "loss": 100.8425,
      "step": 12850
    },
    {
      "epoch": 0.1063820986888365,
      "grad_norm": 1191.3555908203125,
      "learning_rate": 9.99953211295153e-06,
      "loss": 147.7438,
      "step": 12860
    },
    {
      "epoch": 0.10646482193820574,
      "grad_norm": 1642.895263671875,
      "learning_rate": 9.999512620046523e-06,
      "loss": 170.0533,
      "step": 12870
    },
    {
      "epoch": 0.10654754518757496,
      "grad_norm": 1362.4990234375,
      "learning_rate": 9.999492729353238e-06,
      "loss": 177.1871,
      "step": 12880
    },
    {
      "epoch": 0.1066302684369442,
      "grad_norm": 879.8888549804688,
      "learning_rate": 9.999472440873261e-06,
      "loss": 160.7778,
      "step": 12890
    },
    {
      "epoch": 0.10671299168631344,
      "grad_norm": 1243.377685546875,
      "learning_rate": 9.999451754608208e-06,
      "loss": 150.4305,
      "step": 12900
    },
    {
      "epoch": 0.10679571493568267,
      "grad_norm": 984.1234130859375,
      "learning_rate": 9.999430670559723e-06,
      "loss": 139.6723,
      "step": 12910
    },
    {
      "epoch": 0.10687843818505191,
      "grad_norm": 852.8709716796875,
      "learning_rate": 9.999409188729484e-06,
      "loss": 134.8206,
      "step": 12920
    },
    {
      "epoch": 0.10696116143442115,
      "grad_norm": 886.4750366210938,
      "learning_rate": 9.999387309119198e-06,
      "loss": 150.4601,
      "step": 12930
    },
    {
      "epoch": 0.10704388468379038,
      "grad_norm": 1050.5213623046875,
      "learning_rate": 9.999365031730609e-06,
      "loss": 162.7591,
      "step": 12940
    },
    {
      "epoch": 0.10712660793315962,
      "grad_norm": 1679.367919921875,
      "learning_rate": 9.99934235656549e-06,
      "loss": 146.1199,
      "step": 12950
    },
    {
      "epoch": 0.10720933118252884,
      "grad_norm": 1546.12939453125,
      "learning_rate": 9.999319283625641e-06,
      "loss": 207.9114,
      "step": 12960
    },
    {
      "epoch": 0.10729205443189808,
      "grad_norm": 856.313720703125,
      "learning_rate": 9.999295812912902e-06,
      "loss": 168.0912,
      "step": 12970
    },
    {
      "epoch": 0.10737477768126732,
      "grad_norm": 1466.169677734375,
      "learning_rate": 9.999271944429139e-06,
      "loss": 202.6795,
      "step": 12980
    },
    {
      "epoch": 0.10745750093063655,
      "grad_norm": 1105.8492431640625,
      "learning_rate": 9.99924767817625e-06,
      "loss": 139.1426,
      "step": 12990
    },
    {
      "epoch": 0.10754022418000579,
      "grad_norm": 1072.4610595703125,
      "learning_rate": 9.999223014156167e-06,
      "loss": 196.2584,
      "step": 13000
    },
    {
      "epoch": 0.10762294742937503,
      "grad_norm": 820.6765747070312,
      "learning_rate": 9.999197952370851e-06,
      "loss": 177.7073,
      "step": 13010
    },
    {
      "epoch": 0.10770567067874426,
      "grad_norm": 788.7813720703125,
      "learning_rate": 9.9991724928223e-06,
      "loss": 157.0522,
      "step": 13020
    },
    {
      "epoch": 0.1077883939281135,
      "grad_norm": 1179.322265625,
      "learning_rate": 9.999146635512535e-06,
      "loss": 189.2783,
      "step": 13030
    },
    {
      "epoch": 0.10787111717748273,
      "grad_norm": 1035.255615234375,
      "learning_rate": 9.999120380443614e-06,
      "loss": 145.2693,
      "step": 13040
    },
    {
      "epoch": 0.10795384042685197,
      "grad_norm": 1015.682861328125,
      "learning_rate": 9.99909372761763e-06,
      "loss": 143.6347,
      "step": 13050
    },
    {
      "epoch": 0.1080365636762212,
      "grad_norm": 2936.619873046875,
      "learning_rate": 9.9990666770367e-06,
      "loss": 179.5399,
      "step": 13060
    },
    {
      "epoch": 0.10811928692559043,
      "grad_norm": 1349.863037109375,
      "learning_rate": 9.999039228702975e-06,
      "loss": 176.5338,
      "step": 13070
    },
    {
      "epoch": 0.10820201017495967,
      "grad_norm": 1154.5147705078125,
      "learning_rate": 9.999011382618644e-06,
      "loss": 166.4444,
      "step": 13080
    },
    {
      "epoch": 0.10828473342432891,
      "grad_norm": 1495.1844482421875,
      "learning_rate": 9.998983138785919e-06,
      "loss": 149.649,
      "step": 13090
    },
    {
      "epoch": 0.10836745667369814,
      "grad_norm": 1207.58984375,
      "learning_rate": 9.998954497207045e-06,
      "loss": 149.596,
      "step": 13100
    },
    {
      "epoch": 0.10845017992306738,
      "grad_norm": 1018.6047973632812,
      "learning_rate": 9.998925457884307e-06,
      "loss": 154.7165,
      "step": 13110
    },
    {
      "epoch": 0.1085329031724366,
      "grad_norm": 1084.2874755859375,
      "learning_rate": 9.99889602082001e-06,
      "loss": 145.5709,
      "step": 13120
    },
    {
      "epoch": 0.10861562642180585,
      "grad_norm": 1162.242919921875,
      "learning_rate": 9.998866186016501e-06,
      "loss": 168.6215,
      "step": 13130
    },
    {
      "epoch": 0.10869834967117509,
      "grad_norm": 1217.2177734375,
      "learning_rate": 9.99883595347615e-06,
      "loss": 150.9769,
      "step": 13140
    },
    {
      "epoch": 0.10878107292054431,
      "grad_norm": 884.3536376953125,
      "learning_rate": 9.998805323201364e-06,
      "loss": 161.5837,
      "step": 13150
    },
    {
      "epoch": 0.10886379616991355,
      "grad_norm": 1069.2374267578125,
      "learning_rate": 9.998774295194579e-06,
      "loss": 167.3784,
      "step": 13160
    },
    {
      "epoch": 0.1089465194192828,
      "grad_norm": 749.6088256835938,
      "learning_rate": 9.998742869458264e-06,
      "loss": 115.7084,
      "step": 13170
    },
    {
      "epoch": 0.10902924266865202,
      "grad_norm": 703.9533081054688,
      "learning_rate": 9.998711045994922e-06,
      "loss": 162.0743,
      "step": 13180
    },
    {
      "epoch": 0.10911196591802126,
      "grad_norm": 1521.7666015625,
      "learning_rate": 9.998678824807082e-06,
      "loss": 189.686,
      "step": 13190
    },
    {
      "epoch": 0.1091946891673905,
      "grad_norm": 1742.9715576171875,
      "learning_rate": 9.99864620589731e-06,
      "loss": 163.9251,
      "step": 13200
    },
    {
      "epoch": 0.10927741241675973,
      "grad_norm": 999.13623046875,
      "learning_rate": 9.998613189268197e-06,
      "loss": 142.0809,
      "step": 13210
    },
    {
      "epoch": 0.10936013566612897,
      "grad_norm": 1409.1549072265625,
      "learning_rate": 9.998579774922377e-06,
      "loss": 149.1874,
      "step": 13220
    },
    {
      "epoch": 0.1094428589154982,
      "grad_norm": 675.3158569335938,
      "learning_rate": 9.998545962862503e-06,
      "loss": 117.2426,
      "step": 13230
    },
    {
      "epoch": 0.10952558216486744,
      "grad_norm": 1737.0997314453125,
      "learning_rate": 9.998511753091267e-06,
      "loss": 155.6872,
      "step": 13240
    },
    {
      "epoch": 0.10960830541423668,
      "grad_norm": 4492.49853515625,
      "learning_rate": 9.998477145611389e-06,
      "loss": 195.1054,
      "step": 13250
    },
    {
      "epoch": 0.1096910286636059,
      "grad_norm": 1431.2005615234375,
      "learning_rate": 9.998442140425625e-06,
      "loss": 189.3633,
      "step": 13260
    },
    {
      "epoch": 0.10977375191297514,
      "grad_norm": 1214.457763671875,
      "learning_rate": 9.998406737536761e-06,
      "loss": 167.9152,
      "step": 13270
    },
    {
      "epoch": 0.10985647516234438,
      "grad_norm": 553.1201171875,
      "learning_rate": 9.998370936947614e-06,
      "loss": 154.5592,
      "step": 13280
    },
    {
      "epoch": 0.10993919841171361,
      "grad_norm": 1713.546875,
      "learning_rate": 9.998334738661028e-06,
      "loss": 178.8647,
      "step": 13290
    },
    {
      "epoch": 0.11002192166108285,
      "grad_norm": 1338.5965576171875,
      "learning_rate": 9.998298142679888e-06,
      "loss": 209.809,
      "step": 13300
    },
    {
      "epoch": 0.11010464491045208,
      "grad_norm": 1447.3448486328125,
      "learning_rate": 9.998261149007104e-06,
      "loss": 151.2987,
      "step": 13310
    },
    {
      "epoch": 0.11018736815982132,
      "grad_norm": 1176.512939453125,
      "learning_rate": 9.998223757645618e-06,
      "loss": 154.0995,
      "step": 13320
    },
    {
      "epoch": 0.11027009140919056,
      "grad_norm": 1491.28466796875,
      "learning_rate": 9.998185968598407e-06,
      "loss": 172.6219,
      "step": 13330
    },
    {
      "epoch": 0.11035281465855978,
      "grad_norm": 872.3587036132812,
      "learning_rate": 9.998147781868477e-06,
      "loss": 136.3148,
      "step": 13340
    },
    {
      "epoch": 0.11043553790792902,
      "grad_norm": 1718.0472412109375,
      "learning_rate": 9.998109197458865e-06,
      "loss": 147.2434,
      "step": 13350
    },
    {
      "epoch": 0.11051826115729826,
      "grad_norm": 1724.40966796875,
      "learning_rate": 9.998070215372645e-06,
      "loss": 155.5677,
      "step": 13360
    },
    {
      "epoch": 0.11060098440666749,
      "grad_norm": 1238.861572265625,
      "learning_rate": 9.998030835612914e-06,
      "loss": 177.9599,
      "step": 13370
    },
    {
      "epoch": 0.11068370765603673,
      "grad_norm": 1163.747802734375,
      "learning_rate": 9.997991058182807e-06,
      "loss": 159.718,
      "step": 13380
    },
    {
      "epoch": 0.11076643090540596,
      "grad_norm": 1202.32763671875,
      "learning_rate": 9.997950883085492e-06,
      "loss": 161.9838,
      "step": 13390
    },
    {
      "epoch": 0.1108491541547752,
      "grad_norm": 1815.2415771484375,
      "learning_rate": 9.99791031032416e-06,
      "loss": 187.6479,
      "step": 13400
    },
    {
      "epoch": 0.11093187740414444,
      "grad_norm": 1177.880126953125,
      "learning_rate": 9.997869339902043e-06,
      "loss": 180.2671,
      "step": 13410
    },
    {
      "epoch": 0.11101460065351366,
      "grad_norm": 1133.1861572265625,
      "learning_rate": 9.9978279718224e-06,
      "loss": 161.116,
      "step": 13420
    },
    {
      "epoch": 0.1110973239028829,
      "grad_norm": 1010.6656494140625,
      "learning_rate": 9.99778620608852e-06,
      "loss": 178.742,
      "step": 13430
    },
    {
      "epoch": 0.11118004715225215,
      "grad_norm": 1107.10986328125,
      "learning_rate": 9.997744042703731e-06,
      "loss": 139.692,
      "step": 13440
    },
    {
      "epoch": 0.11126277040162137,
      "grad_norm": 1005.2382202148438,
      "learning_rate": 9.997701481671384e-06,
      "loss": 178.2121,
      "step": 13450
    },
    {
      "epoch": 0.11134549365099061,
      "grad_norm": 2284.540283203125,
      "learning_rate": 9.997658522994867e-06,
      "loss": 156.5188,
      "step": 13460
    },
    {
      "epoch": 0.11142821690035985,
      "grad_norm": 5658.24658203125,
      "learning_rate": 9.997615166677597e-06,
      "loss": 146.769,
      "step": 13470
    },
    {
      "epoch": 0.11151094014972908,
      "grad_norm": 457.560302734375,
      "learning_rate": 9.997571412723024e-06,
      "loss": 119.6845,
      "step": 13480
    },
    {
      "epoch": 0.11159366339909832,
      "grad_norm": 1155.69677734375,
      "learning_rate": 9.99752726113463e-06,
      "loss": 126.1926,
      "step": 13490
    },
    {
      "epoch": 0.11167638664846755,
      "grad_norm": 1057.7230224609375,
      "learning_rate": 9.997482711915926e-06,
      "loss": 145.3562,
      "step": 13500
    },
    {
      "epoch": 0.11175910989783679,
      "grad_norm": 1107.236328125,
      "learning_rate": 9.99743776507046e-06,
      "loss": 189.204,
      "step": 13510
    },
    {
      "epoch": 0.11184183314720603,
      "grad_norm": 1068.25634765625,
      "learning_rate": 9.997392420601804e-06,
      "loss": 135.3788,
      "step": 13520
    },
    {
      "epoch": 0.11192455639657525,
      "grad_norm": 875.785888671875,
      "learning_rate": 9.99734667851357e-06,
      "loss": 152.927,
      "step": 13530
    },
    {
      "epoch": 0.1120072796459445,
      "grad_norm": 1772.181396484375,
      "learning_rate": 9.997300538809394e-06,
      "loss": 222.4996,
      "step": 13540
    },
    {
      "epoch": 0.11209000289531373,
      "grad_norm": 1264.271240234375,
      "learning_rate": 9.99725400149295e-06,
      "loss": 145.0414,
      "step": 13550
    },
    {
      "epoch": 0.11217272614468296,
      "grad_norm": 2198.46630859375,
      "learning_rate": 9.997207066567939e-06,
      "loss": 194.5429,
      "step": 13560
    },
    {
      "epoch": 0.1122554493940522,
      "grad_norm": 1130.449951171875,
      "learning_rate": 9.997159734038096e-06,
      "loss": 166.257,
      "step": 13570
    },
    {
      "epoch": 0.11233817264342143,
      "grad_norm": 1828.01513671875,
      "learning_rate": 9.997112003907186e-06,
      "loss": 152.8911,
      "step": 13580
    },
    {
      "epoch": 0.11242089589279067,
      "grad_norm": 1196.669677734375,
      "learning_rate": 9.997063876179007e-06,
      "loss": 129.7313,
      "step": 13590
    },
    {
      "epoch": 0.11250361914215991,
      "grad_norm": 1254.476806640625,
      "learning_rate": 9.997015350857391e-06,
      "loss": 169.0213,
      "step": 13600
    },
    {
      "epoch": 0.11258634239152913,
      "grad_norm": 1430.6507568359375,
      "learning_rate": 9.996966427946195e-06,
      "loss": 150.1627,
      "step": 13610
    },
    {
      "epoch": 0.11266906564089837,
      "grad_norm": 906.8787231445312,
      "learning_rate": 9.996917107449313e-06,
      "loss": 174.3134,
      "step": 13620
    },
    {
      "epoch": 0.11275178889026762,
      "grad_norm": 812.057373046875,
      "learning_rate": 9.99686738937067e-06,
      "loss": 138.5191,
      "step": 13630
    },
    {
      "epoch": 0.11283451213963684,
      "grad_norm": 1879.1136474609375,
      "learning_rate": 9.996817273714222e-06,
      "loss": 174.5974,
      "step": 13640
    },
    {
      "epoch": 0.11291723538900608,
      "grad_norm": 945.3467407226562,
      "learning_rate": 9.996766760483955e-06,
      "loss": 161.973,
      "step": 13650
    },
    {
      "epoch": 0.11299995863837531,
      "grad_norm": 862.3170776367188,
      "learning_rate": 9.996715849683889e-06,
      "loss": 137.5633,
      "step": 13660
    },
    {
      "epoch": 0.11308268188774455,
      "grad_norm": 690.8417358398438,
      "learning_rate": 9.996664541318076e-06,
      "loss": 141.2179,
      "step": 13670
    },
    {
      "epoch": 0.11316540513711379,
      "grad_norm": 972.900634765625,
      "learning_rate": 9.996612835390596e-06,
      "loss": 115.8736,
      "step": 13680
    },
    {
      "epoch": 0.11324812838648302,
      "grad_norm": 1904.7318115234375,
      "learning_rate": 9.996560731905565e-06,
      "loss": 154.7887,
      "step": 13690
    },
    {
      "epoch": 0.11333085163585226,
      "grad_norm": 851.8038940429688,
      "learning_rate": 9.996508230867126e-06,
      "loss": 137.4024,
      "step": 13700
    },
    {
      "epoch": 0.1134135748852215,
      "grad_norm": 1190.90380859375,
      "learning_rate": 9.996455332279458e-06,
      "loss": 153.5202,
      "step": 13710
    },
    {
      "epoch": 0.11349629813459072,
      "grad_norm": 2241.255126953125,
      "learning_rate": 9.99640203614677e-06,
      "loss": 173.7966,
      "step": 13720
    },
    {
      "epoch": 0.11357902138395996,
      "grad_norm": 1084.945068359375,
      "learning_rate": 9.996348342473304e-06,
      "loss": 159.4762,
      "step": 13730
    },
    {
      "epoch": 0.1136617446333292,
      "grad_norm": 1753.5206298828125,
      "learning_rate": 9.99629425126333e-06,
      "loss": 218.3281,
      "step": 13740
    },
    {
      "epoch": 0.11374446788269843,
      "grad_norm": 802.15576171875,
      "learning_rate": 9.996239762521152e-06,
      "loss": 153.0149,
      "step": 13750
    },
    {
      "epoch": 0.11382719113206767,
      "grad_norm": 2257.760986328125,
      "learning_rate": 9.996184876251105e-06,
      "loss": 134.656,
      "step": 13760
    },
    {
      "epoch": 0.1139099143814369,
      "grad_norm": 1485.7083740234375,
      "learning_rate": 9.996129592457558e-06,
      "loss": 119.6472,
      "step": 13770
    },
    {
      "epoch": 0.11399263763080614,
      "grad_norm": 1323.489990234375,
      "learning_rate": 9.996073911144907e-06,
      "loss": 135.8627,
      "step": 13780
    },
    {
      "epoch": 0.11407536088017538,
      "grad_norm": 1439.948974609375,
      "learning_rate": 9.996017832317583e-06,
      "loss": 105.2017,
      "step": 13790
    },
    {
      "epoch": 0.1141580841295446,
      "grad_norm": 755.0234375,
      "learning_rate": 9.995961355980052e-06,
      "loss": 119.7319,
      "step": 13800
    },
    {
      "epoch": 0.11424080737891384,
      "grad_norm": 1265.96826171875,
      "learning_rate": 9.995904482136803e-06,
      "loss": 144.368,
      "step": 13810
    },
    {
      "epoch": 0.11432353062828308,
      "grad_norm": 1014.328857421875,
      "learning_rate": 9.99584721079236e-06,
      "loss": 180.5596,
      "step": 13820
    },
    {
      "epoch": 0.11440625387765231,
      "grad_norm": 1355.7919921875,
      "learning_rate": 9.995789541951287e-06,
      "loss": 169.1609,
      "step": 13830
    },
    {
      "epoch": 0.11448897712702155,
      "grad_norm": 1709.610107421875,
      "learning_rate": 9.995731475618163e-06,
      "loss": 152.0147,
      "step": 13840
    },
    {
      "epoch": 0.11457170037639078,
      "grad_norm": 1426.723388671875,
      "learning_rate": 9.995673011797615e-06,
      "loss": 142.7122,
      "step": 13850
    },
    {
      "epoch": 0.11465442362576002,
      "grad_norm": 3322.473876953125,
      "learning_rate": 9.995614150494293e-06,
      "loss": 192.3159,
      "step": 13860
    },
    {
      "epoch": 0.11473714687512926,
      "grad_norm": 852.6958618164062,
      "learning_rate": 9.995554891712879e-06,
      "loss": 221.4455,
      "step": 13870
    },
    {
      "epoch": 0.11481987012449849,
      "grad_norm": 679.8392944335938,
      "learning_rate": 9.995495235458087e-06,
      "loss": 205.1969,
      "step": 13880
    },
    {
      "epoch": 0.11490259337386773,
      "grad_norm": 1818.237548828125,
      "learning_rate": 9.99543518173467e-06,
      "loss": 157.8227,
      "step": 13890
    },
    {
      "epoch": 0.11498531662323697,
      "grad_norm": 911.2511596679688,
      "learning_rate": 9.995374730547397e-06,
      "loss": 213.6541,
      "step": 13900
    },
    {
      "epoch": 0.11506803987260619,
      "grad_norm": 1822.821044921875,
      "learning_rate": 9.995313881901085e-06,
      "loss": 198.7188,
      "step": 13910
    },
    {
      "epoch": 0.11515076312197543,
      "grad_norm": 653.8102416992188,
      "learning_rate": 9.995252635800572e-06,
      "loss": 127.7723,
      "step": 13920
    },
    {
      "epoch": 0.11523348637134467,
      "grad_norm": 1713.5816650390625,
      "learning_rate": 9.995190992250732e-06,
      "loss": 225.4239,
      "step": 13930
    },
    {
      "epoch": 0.1153162096207139,
      "grad_norm": 1139.3336181640625,
      "learning_rate": 9.995128951256469e-06,
      "loss": 140.1807,
      "step": 13940
    },
    {
      "epoch": 0.11539893287008314,
      "grad_norm": 1414.932861328125,
      "learning_rate": 9.99506651282272e-06,
      "loss": 161.69,
      "step": 13950
    },
    {
      "epoch": 0.11548165611945237,
      "grad_norm": 1176.623291015625,
      "learning_rate": 9.995003676954454e-06,
      "loss": 151.0156,
      "step": 13960
    },
    {
      "epoch": 0.11556437936882161,
      "grad_norm": 784.6156616210938,
      "learning_rate": 9.994940443656668e-06,
      "loss": 198.3028,
      "step": 13970
    },
    {
      "epoch": 0.11564710261819085,
      "grad_norm": 862.885986328125,
      "learning_rate": 9.994876812934395e-06,
      "loss": 153.6012,
      "step": 13980
    },
    {
      "epoch": 0.11572982586756007,
      "grad_norm": 2193.662109375,
      "learning_rate": 9.994812784792698e-06,
      "loss": 165.4299,
      "step": 13990
    },
    {
      "epoch": 0.11581254911692931,
      "grad_norm": 1005.8052978515625,
      "learning_rate": 9.99474835923667e-06,
      "loss": 117.067,
      "step": 14000
    },
    {
      "epoch": 0.11589527236629855,
      "grad_norm": 870.8054809570312,
      "learning_rate": 9.994683536271437e-06,
      "loss": 177.513,
      "step": 14010
    },
    {
      "epoch": 0.11597799561566778,
      "grad_norm": 1938.4117431640625,
      "learning_rate": 9.994618315902161e-06,
      "loss": 147.8295,
      "step": 14020
    },
    {
      "epoch": 0.11606071886503702,
      "grad_norm": 1145.3619384765625,
      "learning_rate": 9.994552698134023e-06,
      "loss": 126.2492,
      "step": 14030
    },
    {
      "epoch": 0.11614344211440625,
      "grad_norm": 704.4757080078125,
      "learning_rate": 9.994486682972253e-06,
      "loss": 183.3489,
      "step": 14040
    },
    {
      "epoch": 0.11622616536377549,
      "grad_norm": 799.8057250976562,
      "learning_rate": 9.994420270422096e-06,
      "loss": 155.8286,
      "step": 14050
    },
    {
      "epoch": 0.11630888861314473,
      "grad_norm": 2168.39794921875,
      "learning_rate": 9.994353460488842e-06,
      "loss": 165.6206,
      "step": 14060
    },
    {
      "epoch": 0.11639161186251396,
      "grad_norm": 1048.3438720703125,
      "learning_rate": 9.994286253177803e-06,
      "loss": 196.4472,
      "step": 14070
    },
    {
      "epoch": 0.1164743351118832,
      "grad_norm": 1240.358642578125,
      "learning_rate": 9.994218648494327e-06,
      "loss": 169.1644,
      "step": 14080
    },
    {
      "epoch": 0.11655705836125244,
      "grad_norm": 1450.99169921875,
      "learning_rate": 9.994150646443793e-06,
      "loss": 119.286,
      "step": 14090
    },
    {
      "epoch": 0.11663978161062166,
      "grad_norm": 1026.7149658203125,
      "learning_rate": 9.994082247031613e-06,
      "loss": 166.7578,
      "step": 14100
    },
    {
      "epoch": 0.1167225048599909,
      "grad_norm": 1081.6656494140625,
      "learning_rate": 9.99401345026323e-06,
      "loss": 141.0757,
      "step": 14110
    },
    {
      "epoch": 0.11680522810936013,
      "grad_norm": 746.9979248046875,
      "learning_rate": 9.993944256144115e-06,
      "loss": 124.9759,
      "step": 14120
    },
    {
      "epoch": 0.11688795135872937,
      "grad_norm": 891.9210815429688,
      "learning_rate": 9.993874664679774e-06,
      "loss": 150.3685,
      "step": 14130
    },
    {
      "epoch": 0.11697067460809861,
      "grad_norm": 1401.4002685546875,
      "learning_rate": 9.993804675875744e-06,
      "loss": 168.8493,
      "step": 14140
    },
    {
      "epoch": 0.11705339785746784,
      "grad_norm": 1588.7640380859375,
      "learning_rate": 9.993734289737596e-06,
      "loss": 141.4464,
      "step": 14150
    },
    {
      "epoch": 0.11713612110683708,
      "grad_norm": 1859.66552734375,
      "learning_rate": 9.993663506270928e-06,
      "loss": 162.024,
      "step": 14160
    },
    {
      "epoch": 0.11721884435620632,
      "grad_norm": 1133.1839599609375,
      "learning_rate": 9.993592325481373e-06,
      "loss": 166.6096,
      "step": 14170
    },
    {
      "epoch": 0.11730156760557554,
      "grad_norm": 1811.849365234375,
      "learning_rate": 9.993520747374594e-06,
      "loss": 127.2197,
      "step": 14180
    },
    {
      "epoch": 0.11738429085494478,
      "grad_norm": 909.2362060546875,
      "learning_rate": 9.993448771956285e-06,
      "loss": 189.4919,
      "step": 14190
    },
    {
      "epoch": 0.11746701410431402,
      "grad_norm": 1350.44140625,
      "learning_rate": 9.993376399232175e-06,
      "loss": 142.4382,
      "step": 14200
    },
    {
      "epoch": 0.11754973735368325,
      "grad_norm": 1765.2679443359375,
      "learning_rate": 9.993303629208023e-06,
      "loss": 148.8411,
      "step": 14210
    },
    {
      "epoch": 0.11763246060305249,
      "grad_norm": 2343.818359375,
      "learning_rate": 9.993230461889616e-06,
      "loss": 212.7168,
      "step": 14220
    },
    {
      "epoch": 0.11771518385242172,
      "grad_norm": 683.275146484375,
      "learning_rate": 9.993156897282776e-06,
      "loss": 148.4446,
      "step": 14230
    },
    {
      "epoch": 0.11779790710179096,
      "grad_norm": 998.6349487304688,
      "learning_rate": 9.99308293539336e-06,
      "loss": 117.4103,
      "step": 14240
    },
    {
      "epoch": 0.1178806303511602,
      "grad_norm": 823.5191040039062,
      "learning_rate": 9.993008576227248e-06,
      "loss": 130.8048,
      "step": 14250
    },
    {
      "epoch": 0.11796335360052942,
      "grad_norm": 1420.8035888671875,
      "learning_rate": 9.992933819790358e-06,
      "loss": 163.5295,
      "step": 14260
    },
    {
      "epoch": 0.11804607684989867,
      "grad_norm": 1167.037109375,
      "learning_rate": 9.992858666088638e-06,
      "loss": 164.6194,
      "step": 14270
    },
    {
      "epoch": 0.1181288000992679,
      "grad_norm": 1515.985107421875,
      "learning_rate": 9.992783115128072e-06,
      "loss": 163.406,
      "step": 14280
    },
    {
      "epoch": 0.11821152334863713,
      "grad_norm": 1549.8917236328125,
      "learning_rate": 9.992707166914662e-06,
      "loss": 168.7726,
      "step": 14290
    },
    {
      "epoch": 0.11829424659800637,
      "grad_norm": 1190.7861328125,
      "learning_rate": 9.992630821454458e-06,
      "loss": 140.9276,
      "step": 14300
    },
    {
      "epoch": 0.1183769698473756,
      "grad_norm": 1568.7037353515625,
      "learning_rate": 9.992554078753534e-06,
      "loss": 147.5554,
      "step": 14310
    },
    {
      "epoch": 0.11845969309674484,
      "grad_norm": 1000.02880859375,
      "learning_rate": 9.992476938817994e-06,
      "loss": 180.1213,
      "step": 14320
    },
    {
      "epoch": 0.11854241634611408,
      "grad_norm": 1296.5947265625,
      "learning_rate": 9.992399401653976e-06,
      "loss": 137.781,
      "step": 14330
    },
    {
      "epoch": 0.1186251395954833,
      "grad_norm": 1144.0504150390625,
      "learning_rate": 9.99232146726765e-06,
      "loss": 140.0204,
      "step": 14340
    },
    {
      "epoch": 0.11870786284485255,
      "grad_norm": 1281.876708984375,
      "learning_rate": 9.992243135665217e-06,
      "loss": 154.8919,
      "step": 14350
    },
    {
      "epoch": 0.11879058609422179,
      "grad_norm": 871.8610229492188,
      "learning_rate": 9.992164406852908e-06,
      "loss": 186.1516,
      "step": 14360
    },
    {
      "epoch": 0.11887330934359101,
      "grad_norm": 1534.3536376953125,
      "learning_rate": 9.992085280836988e-06,
      "loss": 160.6092,
      "step": 14370
    },
    {
      "epoch": 0.11895603259296025,
      "grad_norm": 988.0948486328125,
      "learning_rate": 9.992005757623753e-06,
      "loss": 203.4977,
      "step": 14380
    },
    {
      "epoch": 0.11903875584232948,
      "grad_norm": 1357.1668701171875,
      "learning_rate": 9.991925837219532e-06,
      "loss": 160.5042,
      "step": 14390
    },
    {
      "epoch": 0.11912147909169872,
      "grad_norm": 636.4329223632812,
      "learning_rate": 9.991845519630679e-06,
      "loss": 137.9073,
      "step": 14400
    },
    {
      "epoch": 0.11920420234106796,
      "grad_norm": 1027.252197265625,
      "learning_rate": 9.991764804863588e-06,
      "loss": 122.9028,
      "step": 14410
    },
    {
      "epoch": 0.11928692559043719,
      "grad_norm": 2356.058837890625,
      "learning_rate": 9.991683692924682e-06,
      "loss": 155.1582,
      "step": 14420
    },
    {
      "epoch": 0.11936964883980643,
      "grad_norm": 743.142578125,
      "learning_rate": 9.991602183820412e-06,
      "loss": 142.9998,
      "step": 14430
    },
    {
      "epoch": 0.11945237208917567,
      "grad_norm": 1039.77978515625,
      "learning_rate": 9.991520277557266e-06,
      "loss": 156.9646,
      "step": 14440
    },
    {
      "epoch": 0.1195350953385449,
      "grad_norm": 1624.250732421875,
      "learning_rate": 9.991437974141759e-06,
      "loss": 165.5059,
      "step": 14450
    },
    {
      "epoch": 0.11961781858791413,
      "grad_norm": 589.4429321289062,
      "learning_rate": 9.99135527358044e-06,
      "loss": 119.63,
      "step": 14460
    },
    {
      "epoch": 0.11970054183728338,
      "grad_norm": 1199.8302001953125,
      "learning_rate": 9.991272175879888e-06,
      "loss": 175.7935,
      "step": 14470
    },
    {
      "epoch": 0.1197832650866526,
      "grad_norm": 1233.6771240234375,
      "learning_rate": 9.991188681046718e-06,
      "loss": 192.3081,
      "step": 14480
    },
    {
      "epoch": 0.11986598833602184,
      "grad_norm": 1209.5980224609375,
      "learning_rate": 9.991104789087568e-06,
      "loss": 139.1357,
      "step": 14490
    },
    {
      "epoch": 0.11994871158539107,
      "grad_norm": 1937.0008544921875,
      "learning_rate": 9.991020500009118e-06,
      "loss": 138.8762,
      "step": 14500
    },
    {
      "epoch": 0.12003143483476031,
      "grad_norm": 545.6674194335938,
      "learning_rate": 9.990935813818073e-06,
      "loss": 139.6014,
      "step": 14510
    },
    {
      "epoch": 0.12011415808412955,
      "grad_norm": 650.5000610351562,
      "learning_rate": 9.99085073052117e-06,
      "loss": 195.5592,
      "step": 14520
    },
    {
      "epoch": 0.12019688133349878,
      "grad_norm": 566.500732421875,
      "learning_rate": 9.990765250125179e-06,
      "loss": 154.2413,
      "step": 14530
    },
    {
      "epoch": 0.12027960458286802,
      "grad_norm": 1750.378662109375,
      "learning_rate": 9.990679372636902e-06,
      "loss": 161.8778,
      "step": 14540
    },
    {
      "epoch": 0.12036232783223726,
      "grad_norm": 1349.2432861328125,
      "learning_rate": 9.99059309806317e-06,
      "loss": 172.9573,
      "step": 14550
    },
    {
      "epoch": 0.12044505108160648,
      "grad_norm": 975.1179809570312,
      "learning_rate": 9.990506426410851e-06,
      "loss": 128.1013,
      "step": 14560
    },
    {
      "epoch": 0.12052777433097572,
      "grad_norm": 3406.892333984375,
      "learning_rate": 9.990419357686839e-06,
      "loss": 177.5993,
      "step": 14570
    },
    {
      "epoch": 0.12061049758034495,
      "grad_norm": 428.1646423339844,
      "learning_rate": 9.99033189189806e-06,
      "loss": 128.9519,
      "step": 14580
    },
    {
      "epoch": 0.12069322082971419,
      "grad_norm": 996.7293701171875,
      "learning_rate": 9.990244029051475e-06,
      "loss": 149.3833,
      "step": 14590
    },
    {
      "epoch": 0.12077594407908343,
      "grad_norm": 1458.9307861328125,
      "learning_rate": 9.990155769154077e-06,
      "loss": 160.3518,
      "step": 14600
    },
    {
      "epoch": 0.12085866732845266,
      "grad_norm": 714.3155517578125,
      "learning_rate": 9.990067112212884e-06,
      "loss": 128.5733,
      "step": 14610
    },
    {
      "epoch": 0.1209413905778219,
      "grad_norm": 629.7801513671875,
      "learning_rate": 9.989978058234952e-06,
      "loss": 141.1231,
      "step": 14620
    },
    {
      "epoch": 0.12102411382719114,
      "grad_norm": 1056.1544189453125,
      "learning_rate": 9.989888607227369e-06,
      "loss": 173.8705,
      "step": 14630
    },
    {
      "epoch": 0.12110683707656036,
      "grad_norm": 1272.472412109375,
      "learning_rate": 9.989798759197247e-06,
      "loss": 146.9385,
      "step": 14640
    },
    {
      "epoch": 0.1211895603259296,
      "grad_norm": 1054.3629150390625,
      "learning_rate": 9.989708514151739e-06,
      "loss": 164.1719,
      "step": 14650
    },
    {
      "epoch": 0.12127228357529883,
      "grad_norm": 659.7613525390625,
      "learning_rate": 9.989617872098026e-06,
      "loss": 149.6539,
      "step": 14660
    },
    {
      "epoch": 0.12135500682466807,
      "grad_norm": 815.8479614257812,
      "learning_rate": 9.989526833043316e-06,
      "loss": 140.1702,
      "step": 14670
    },
    {
      "epoch": 0.12143773007403731,
      "grad_norm": 1148.9129638671875,
      "learning_rate": 9.989435396994856e-06,
      "loss": 125.2471,
      "step": 14680
    },
    {
      "epoch": 0.12152045332340654,
      "grad_norm": 3006.224609375,
      "learning_rate": 9.989343563959919e-06,
      "loss": 150.3076,
      "step": 14690
    },
    {
      "epoch": 0.12160317657277578,
      "grad_norm": 1365.89892578125,
      "learning_rate": 9.989251333945813e-06,
      "loss": 179.2145,
      "step": 14700
    },
    {
      "epoch": 0.12168589982214502,
      "grad_norm": 982.8682861328125,
      "learning_rate": 9.989158706959875e-06,
      "loss": 137.9394,
      "step": 14710
    },
    {
      "epoch": 0.12176862307151425,
      "grad_norm": 1103.151123046875,
      "learning_rate": 9.989065683009477e-06,
      "loss": 150.3043,
      "step": 14720
    },
    {
      "epoch": 0.12185134632088349,
      "grad_norm": 1068.439208984375,
      "learning_rate": 9.988972262102018e-06,
      "loss": 115.0475,
      "step": 14730
    },
    {
      "epoch": 0.12193406957025273,
      "grad_norm": 1475.9112548828125,
      "learning_rate": 9.988878444244937e-06,
      "loss": 162.3183,
      "step": 14740
    },
    {
      "epoch": 0.12201679281962195,
      "grad_norm": 1475.7916259765625,
      "learning_rate": 9.988784229445689e-06,
      "loss": 132.9056,
      "step": 14750
    },
    {
      "epoch": 0.12209951606899119,
      "grad_norm": 1580.335205078125,
      "learning_rate": 9.988689617711777e-06,
      "loss": 180.2133,
      "step": 14760
    },
    {
      "epoch": 0.12218223931836042,
      "grad_norm": 1742.8638916015625,
      "learning_rate": 9.988594609050726e-06,
      "loss": 170.6644,
      "step": 14770
    },
    {
      "epoch": 0.12226496256772966,
      "grad_norm": 778.4093017578125,
      "learning_rate": 9.988499203470097e-06,
      "loss": 163.6835,
      "step": 14780
    },
    {
      "epoch": 0.1223476858170989,
      "grad_norm": 908.4758911132812,
      "learning_rate": 9.988403400977482e-06,
      "loss": 143.079,
      "step": 14790
    },
    {
      "epoch": 0.12243040906646813,
      "grad_norm": 1540.624755859375,
      "learning_rate": 9.9883072015805e-06,
      "loss": 160.3763,
      "step": 14800
    },
    {
      "epoch": 0.12251313231583737,
      "grad_norm": 919.294677734375,
      "learning_rate": 9.98821060528681e-06,
      "loss": 165.4283,
      "step": 14810
    },
    {
      "epoch": 0.12259585556520661,
      "grad_norm": 865.2339477539062,
      "learning_rate": 9.988113612104093e-06,
      "loss": 128.951,
      "step": 14820
    },
    {
      "epoch": 0.12267857881457583,
      "grad_norm": 2098.492919921875,
      "learning_rate": 9.988016222040067e-06,
      "loss": 151.1649,
      "step": 14830
    },
    {
      "epoch": 0.12276130206394507,
      "grad_norm": 847.621337890625,
      "learning_rate": 9.987918435102484e-06,
      "loss": 121.6645,
      "step": 14840
    },
    {
      "epoch": 0.1228440253133143,
      "grad_norm": 1472.7208251953125,
      "learning_rate": 9.987820251299121e-06,
      "loss": 140.8588,
      "step": 14850
    },
    {
      "epoch": 0.12292674856268354,
      "grad_norm": 1310.5726318359375,
      "learning_rate": 9.987721670637794e-06,
      "loss": 132.6207,
      "step": 14860
    },
    {
      "epoch": 0.12300947181205278,
      "grad_norm": 788.9578247070312,
      "learning_rate": 9.987622693126342e-06,
      "loss": 139.2334,
      "step": 14870
    },
    {
      "epoch": 0.12309219506142201,
      "grad_norm": 1761.3287353515625,
      "learning_rate": 9.987523318772644e-06,
      "loss": 156.3363,
      "step": 14880
    },
    {
      "epoch": 0.12317491831079125,
      "grad_norm": 1024.44140625,
      "learning_rate": 9.987423547584605e-06,
      "loss": 167.0266,
      "step": 14890
    },
    {
      "epoch": 0.12325764156016049,
      "grad_norm": 891.505126953125,
      "learning_rate": 9.987323379570161e-06,
      "loss": 144.4436,
      "step": 14900
    },
    {
      "epoch": 0.12334036480952972,
      "grad_norm": 1373.43359375,
      "learning_rate": 9.987222814737287e-06,
      "loss": 139.1032,
      "step": 14910
    },
    {
      "epoch": 0.12342308805889896,
      "grad_norm": 2005.6689453125,
      "learning_rate": 9.987121853093982e-06,
      "loss": 179.2018,
      "step": 14920
    },
    {
      "epoch": 0.12350581130826818,
      "grad_norm": 1276.0216064453125,
      "learning_rate": 9.987020494648279e-06,
      "loss": 157.102,
      "step": 14930
    },
    {
      "epoch": 0.12358853455763742,
      "grad_norm": 1294.2474365234375,
      "learning_rate": 9.986918739408241e-06,
      "loss": 176.7196,
      "step": 14940
    },
    {
      "epoch": 0.12367125780700666,
      "grad_norm": 1715.0291748046875,
      "learning_rate": 9.986816587381966e-06,
      "loss": 150.2139,
      "step": 14950
    },
    {
      "epoch": 0.12375398105637589,
      "grad_norm": 1020.9821166992188,
      "learning_rate": 9.986714038577582e-06,
      "loss": 145.5669,
      "step": 14960
    },
    {
      "epoch": 0.12383670430574513,
      "grad_norm": 969.8027954101562,
      "learning_rate": 9.986611093003249e-06,
      "loss": 129.0563,
      "step": 14970
    },
    {
      "epoch": 0.12391942755511437,
      "grad_norm": 1188.594482421875,
      "learning_rate": 9.986507750667157e-06,
      "loss": 130.371,
      "step": 14980
    },
    {
      "epoch": 0.1240021508044836,
      "grad_norm": 2070.416015625,
      "learning_rate": 9.986404011577525e-06,
      "loss": 164.5774,
      "step": 14990
    },
    {
      "epoch": 0.12408487405385284,
      "grad_norm": 1122.52587890625,
      "learning_rate": 9.986299875742612e-06,
      "loss": 187.0694,
      "step": 15000
    },
    {
      "epoch": 0.12416759730322208,
      "grad_norm": 1119.6961669921875,
      "learning_rate": 9.986195343170703e-06,
      "loss": 180.5289,
      "step": 15010
    },
    {
      "epoch": 0.1242503205525913,
      "grad_norm": 1574.2568359375,
      "learning_rate": 9.986090413870114e-06,
      "loss": 144.7522,
      "step": 15020
    },
    {
      "epoch": 0.12433304380196054,
      "grad_norm": 1006.7045288085938,
      "learning_rate": 9.985985087849193e-06,
      "loss": 143.7221,
      "step": 15030
    },
    {
      "epoch": 0.12441576705132977,
      "grad_norm": 1300.2181396484375,
      "learning_rate": 9.98587936511632e-06,
      "loss": 150.1932,
      "step": 15040
    },
    {
      "epoch": 0.12449849030069901,
      "grad_norm": 821.7041625976562,
      "learning_rate": 9.98577324567991e-06,
      "loss": 139.0086,
      "step": 15050
    },
    {
      "epoch": 0.12458121355006825,
      "grad_norm": 830.7269287109375,
      "learning_rate": 9.985666729548404e-06,
      "loss": 146.4651,
      "step": 15060
    },
    {
      "epoch": 0.12466393679943748,
      "grad_norm": 1310.355224609375,
      "learning_rate": 9.985559816730277e-06,
      "loss": 141.5489,
      "step": 15070
    },
    {
      "epoch": 0.12474666004880672,
      "grad_norm": 1190.0335693359375,
      "learning_rate": 9.985452507234037e-06,
      "loss": 144.9001,
      "step": 15080
    },
    {
      "epoch": 0.12482938329817596,
      "grad_norm": 2714.714599609375,
      "learning_rate": 9.98534480106822e-06,
      "loss": 154.9118,
      "step": 15090
    },
    {
      "epoch": 0.12491210654754518,
      "grad_norm": 792.223388671875,
      "learning_rate": 9.985236698241396e-06,
      "loss": 149.7406,
      "step": 15100
    },
    {
      "epoch": 0.12499482979691443,
      "grad_norm": 1287.8345947265625,
      "learning_rate": 9.985128198762168e-06,
      "loss": 171.4261,
      "step": 15110
    },
    {
      "epoch": 0.12507755304628365,
      "grad_norm": 957.4619140625,
      "learning_rate": 9.98501930263917e-06,
      "loss": 234.8733,
      "step": 15120
    },
    {
      "epoch": 0.1251602762956529,
      "grad_norm": 862.0460205078125,
      "learning_rate": 9.984910009881062e-06,
      "loss": 112.6332,
      "step": 15130
    },
    {
      "epoch": 0.12524299954502213,
      "grad_norm": 1100.07080078125,
      "learning_rate": 9.984800320496542e-06,
      "loss": 139.7673,
      "step": 15140
    },
    {
      "epoch": 0.12532572279439136,
      "grad_norm": 1111.9737548828125,
      "learning_rate": 9.984690234494338e-06,
      "loss": 106.7051,
      "step": 15150
    },
    {
      "epoch": 0.12540844604376059,
      "grad_norm": 744.8794555664062,
      "learning_rate": 9.98457975188321e-06,
      "loss": 142.1312,
      "step": 15160
    },
    {
      "epoch": 0.12549116929312984,
      "grad_norm": 923.189697265625,
      "learning_rate": 9.984468872671945e-06,
      "loss": 139.3656,
      "step": 15170
    },
    {
      "epoch": 0.12557389254249907,
      "grad_norm": 1322.687255859375,
      "learning_rate": 9.984357596869369e-06,
      "loss": 148.6495,
      "step": 15180
    },
    {
      "epoch": 0.1256566157918683,
      "grad_norm": 882.3487548828125,
      "learning_rate": 9.984245924484334e-06,
      "loss": 141.7766,
      "step": 15190
    },
    {
      "epoch": 0.12573933904123755,
      "grad_norm": 725.9840698242188,
      "learning_rate": 9.984133855525723e-06,
      "loss": 138.5364,
      "step": 15200
    },
    {
      "epoch": 0.12582206229060677,
      "grad_norm": 1547.522705078125,
      "learning_rate": 9.984021390002458e-06,
      "loss": 136.4458,
      "step": 15210
    },
    {
      "epoch": 0.125904785539976,
      "grad_norm": 1425.56494140625,
      "learning_rate": 9.983908527923486e-06,
      "loss": 222.0387,
      "step": 15220
    },
    {
      "epoch": 0.12598750878934525,
      "grad_norm": 911.7035522460938,
      "learning_rate": 9.983795269297782e-06,
      "loss": 169.7902,
      "step": 15230
    },
    {
      "epoch": 0.12607023203871448,
      "grad_norm": 1066.0264892578125,
      "learning_rate": 9.983681614134363e-06,
      "loss": 122.5573,
      "step": 15240
    },
    {
      "epoch": 0.1261529552880837,
      "grad_norm": 1829.2509765625,
      "learning_rate": 9.98356756244227e-06,
      "loss": 154.7958,
      "step": 15250
    },
    {
      "epoch": 0.12623567853745296,
      "grad_norm": 1402.93408203125,
      "learning_rate": 9.983453114230575e-06,
      "loss": 145.442,
      "step": 15260
    },
    {
      "epoch": 0.1263184017868222,
      "grad_norm": 990.7800903320312,
      "learning_rate": 9.98333826950839e-06,
      "loss": 138.916,
      "step": 15270
    },
    {
      "epoch": 0.12640112503619141,
      "grad_norm": 861.1292724609375,
      "learning_rate": 9.983223028284847e-06,
      "loss": 152.3527,
      "step": 15280
    },
    {
      "epoch": 0.12648384828556067,
      "grad_norm": 887.3511962890625,
      "learning_rate": 9.983107390569118e-06,
      "loss": 129.7973,
      "step": 15290
    },
    {
      "epoch": 0.1265665715349299,
      "grad_norm": 1043.2373046875,
      "learning_rate": 9.982991356370404e-06,
      "loss": 116.1451,
      "step": 15300
    },
    {
      "epoch": 0.12664929478429912,
      "grad_norm": 1244.5079345703125,
      "learning_rate": 9.982874925697937e-06,
      "loss": 221.0664,
      "step": 15310
    },
    {
      "epoch": 0.12673201803366838,
      "grad_norm": 1715.1995849609375,
      "learning_rate": 9.982758098560978e-06,
      "loss": 186.7455,
      "step": 15320
    },
    {
      "epoch": 0.1268147412830376,
      "grad_norm": 679.9988403320312,
      "learning_rate": 9.982640874968827e-06,
      "loss": 171.8672,
      "step": 15330
    },
    {
      "epoch": 0.12689746453240683,
      "grad_norm": 595.40625,
      "learning_rate": 9.98252325493081e-06,
      "loss": 130.3511,
      "step": 15340
    },
    {
      "epoch": 0.12698018778177605,
      "grad_norm": 915.3275146484375,
      "learning_rate": 9.982405238456281e-06,
      "loss": 153.7831,
      "step": 15350
    },
    {
      "epoch": 0.1270629110311453,
      "grad_norm": 1383.0423583984375,
      "learning_rate": 9.982286825554636e-06,
      "loss": 155.1486,
      "step": 15360
    },
    {
      "epoch": 0.12714563428051454,
      "grad_norm": 1527.0670166015625,
      "learning_rate": 9.982168016235292e-06,
      "loss": 235.3831,
      "step": 15370
    },
    {
      "epoch": 0.12722835752988376,
      "grad_norm": 1168.0416259765625,
      "learning_rate": 9.982048810507706e-06,
      "loss": 175.3166,
      "step": 15380
    },
    {
      "epoch": 0.12731108077925302,
      "grad_norm": 1577.9107666015625,
      "learning_rate": 9.98192920838136e-06,
      "loss": 136.4098,
      "step": 15390
    },
    {
      "epoch": 0.12739380402862224,
      "grad_norm": 2239.125244140625,
      "learning_rate": 9.98180920986577e-06,
      "loss": 162.4811,
      "step": 15400
    },
    {
      "epoch": 0.12747652727799147,
      "grad_norm": 1140.6561279296875,
      "learning_rate": 9.981688814970485e-06,
      "loss": 159.3877,
      "step": 15410
    },
    {
      "epoch": 0.12755925052736072,
      "grad_norm": 929.8948974609375,
      "learning_rate": 9.981568023705085e-06,
      "loss": 113.0717,
      "step": 15420
    },
    {
      "epoch": 0.12764197377672995,
      "grad_norm": 1146.56396484375,
      "learning_rate": 9.981446836079178e-06,
      "loss": 121.9914,
      "step": 15430
    },
    {
      "epoch": 0.12772469702609918,
      "grad_norm": 877.8550415039062,
      "learning_rate": 9.981325252102408e-06,
      "loss": 173.6141,
      "step": 15440
    },
    {
      "epoch": 0.12780742027546843,
      "grad_norm": 969.0079956054688,
      "learning_rate": 9.98120327178445e-06,
      "loss": 178.5706,
      "step": 15450
    },
    {
      "epoch": 0.12789014352483766,
      "grad_norm": 1263.2391357421875,
      "learning_rate": 9.981080895135007e-06,
      "loss": 180.7431,
      "step": 15460
    },
    {
      "epoch": 0.12797286677420688,
      "grad_norm": 742.5184326171875,
      "learning_rate": 9.980958122163818e-06,
      "loss": 111.0224,
      "step": 15470
    },
    {
      "epoch": 0.12805559002357614,
      "grad_norm": 1423.19873046875,
      "learning_rate": 9.980834952880652e-06,
      "loss": 128.3473,
      "step": 15480
    },
    {
      "epoch": 0.12813831327294536,
      "grad_norm": 1583.2940673828125,
      "learning_rate": 9.980711387295306e-06,
      "loss": 149.955,
      "step": 15490
    },
    {
      "epoch": 0.1282210365223146,
      "grad_norm": 1052.7265625,
      "learning_rate": 9.980587425417612e-06,
      "loss": 159.9205,
      "step": 15500
    },
    {
      "epoch": 0.12830375977168385,
      "grad_norm": 2138.17724609375,
      "learning_rate": 9.980463067257437e-06,
      "loss": 169.7366,
      "step": 15510
    },
    {
      "epoch": 0.12838648302105307,
      "grad_norm": 1006.1878662109375,
      "learning_rate": 9.980338312824672e-06,
      "loss": 193.1612,
      "step": 15520
    },
    {
      "epoch": 0.1284692062704223,
      "grad_norm": 1047.7593994140625,
      "learning_rate": 9.980213162129244e-06,
      "loss": 175.5892,
      "step": 15530
    },
    {
      "epoch": 0.12855192951979152,
      "grad_norm": 1267.4644775390625,
      "learning_rate": 9.980087615181111e-06,
      "loss": 149.4357,
      "step": 15540
    },
    {
      "epoch": 0.12863465276916078,
      "grad_norm": 1171.0859375,
      "learning_rate": 9.979961671990263e-06,
      "loss": 165.6414,
      "step": 15550
    },
    {
      "epoch": 0.12871737601853,
      "grad_norm": 911.0418701171875,
      "learning_rate": 9.979835332566719e-06,
      "loss": 155.2462,
      "step": 15560
    },
    {
      "epoch": 0.12880009926789923,
      "grad_norm": 1016.1674194335938,
      "learning_rate": 9.97970859692053e-06,
      "loss": 142.4974,
      "step": 15570
    },
    {
      "epoch": 0.12888282251726849,
      "grad_norm": 653.232421875,
      "learning_rate": 9.979581465061784e-06,
      "loss": 155.5012,
      "step": 15580
    },
    {
      "epoch": 0.1289655457666377,
      "grad_norm": 1058.5008544921875,
      "learning_rate": 9.979453937000594e-06,
      "loss": 101.9423,
      "step": 15590
    },
    {
      "epoch": 0.12904826901600694,
      "grad_norm": 820.3455200195312,
      "learning_rate": 9.979326012747106e-06,
      "loss": 117.5258,
      "step": 15600
    },
    {
      "epoch": 0.1291309922653762,
      "grad_norm": 1352.7105712890625,
      "learning_rate": 9.9791976923115e-06,
      "loss": 99.8209,
      "step": 15610
    },
    {
      "epoch": 0.12921371551474542,
      "grad_norm": 1026.9713134765625,
      "learning_rate": 9.979068975703984e-06,
      "loss": 166.7305,
      "step": 15620
    },
    {
      "epoch": 0.12929643876411465,
      "grad_norm": 1245.904296875,
      "learning_rate": 9.978939862934802e-06,
      "loss": 126.3938,
      "step": 15630
    },
    {
      "epoch": 0.1293791620134839,
      "grad_norm": 1658.7640380859375,
      "learning_rate": 9.978810354014223e-06,
      "loss": 135.5493,
      "step": 15640
    },
    {
      "epoch": 0.12946188526285313,
      "grad_norm": 1517.82373046875,
      "learning_rate": 9.978680448952556e-06,
      "loss": 139.2036,
      "step": 15650
    },
    {
      "epoch": 0.12954460851222235,
      "grad_norm": 1112.2469482421875,
      "learning_rate": 9.978550147760133e-06,
      "loss": 127.4167,
      "step": 15660
    },
    {
      "epoch": 0.1296273317615916,
      "grad_norm": 995.6966552734375,
      "learning_rate": 9.978419450447325e-06,
      "loss": 128.1456,
      "step": 15670
    },
    {
      "epoch": 0.12971005501096083,
      "grad_norm": 785.5421142578125,
      "learning_rate": 9.978288357024527e-06,
      "loss": 142.7447,
      "step": 15680
    },
    {
      "epoch": 0.12979277826033006,
      "grad_norm": 3418.008544921875,
      "learning_rate": 9.978156867502173e-06,
      "loss": 161.2918,
      "step": 15690
    },
    {
      "epoch": 0.12987550150969931,
      "grad_norm": 1545.32568359375,
      "learning_rate": 9.978024981890724e-06,
      "loss": 107.0028,
      "step": 15700
    },
    {
      "epoch": 0.12995822475906854,
      "grad_norm": 963.7340087890625,
      "learning_rate": 9.977892700200673e-06,
      "loss": 140.569,
      "step": 15710
    },
    {
      "epoch": 0.13004094800843777,
      "grad_norm": 692.7611694335938,
      "learning_rate": 9.977760022442545e-06,
      "loss": 110.664,
      "step": 15720
    },
    {
      "epoch": 0.130123671257807,
      "grad_norm": 1015.7322998046875,
      "learning_rate": 9.977626948626897e-06,
      "loss": 158.9243,
      "step": 15730
    },
    {
      "epoch": 0.13020639450717625,
      "grad_norm": 1334.6917724609375,
      "learning_rate": 9.977493478764316e-06,
      "loss": 152.3215,
      "step": 15740
    },
    {
      "epoch": 0.13028911775654548,
      "grad_norm": 963.1575927734375,
      "learning_rate": 9.977359612865424e-06,
      "loss": 137.1868,
      "step": 15750
    },
    {
      "epoch": 0.1303718410059147,
      "grad_norm": 1332.0909423828125,
      "learning_rate": 9.97722535094087e-06,
      "loss": 130.2153,
      "step": 15760
    },
    {
      "epoch": 0.13045456425528396,
      "grad_norm": 864.9472045898438,
      "learning_rate": 9.977090693001336e-06,
      "loss": 142.6017,
      "step": 15770
    },
    {
      "epoch": 0.13053728750465318,
      "grad_norm": 1091.3128662109375,
      "learning_rate": 9.976955639057539e-06,
      "loss": 126.0693,
      "step": 15780
    },
    {
      "epoch": 0.1306200107540224,
      "grad_norm": 1137.7115478515625,
      "learning_rate": 9.976820189120223e-06,
      "loss": 147.4185,
      "step": 15790
    },
    {
      "epoch": 0.13070273400339166,
      "grad_norm": 1658.2982177734375,
      "learning_rate": 9.976684343200164e-06,
      "loss": 135.441,
      "step": 15800
    },
    {
      "epoch": 0.1307854572527609,
      "grad_norm": 1823.3642578125,
      "learning_rate": 9.976548101308173e-06,
      "loss": 138.8229,
      "step": 15810
    },
    {
      "epoch": 0.13086818050213012,
      "grad_norm": 1511.6375732421875,
      "learning_rate": 9.976411463455088e-06,
      "loss": 140.3549,
      "step": 15820
    },
    {
      "epoch": 0.13095090375149937,
      "grad_norm": 970.9559326171875,
      "learning_rate": 9.976274429651783e-06,
      "loss": 188.4605,
      "step": 15830
    },
    {
      "epoch": 0.1310336270008686,
      "grad_norm": 1540.7110595703125,
      "learning_rate": 9.976136999909156e-06,
      "loss": 106.2589,
      "step": 15840
    },
    {
      "epoch": 0.13111635025023782,
      "grad_norm": 829.7328491210938,
      "learning_rate": 9.97599917423815e-06,
      "loss": 166.2885,
      "step": 15850
    },
    {
      "epoch": 0.13119907349960708,
      "grad_norm": 0.0,
      "learning_rate": 9.975860952649724e-06,
      "loss": 180.9173,
      "step": 15860
    },
    {
      "epoch": 0.1312817967489763,
      "grad_norm": 1011.6902465820312,
      "learning_rate": 9.975722335154876e-06,
      "loss": 161.2201,
      "step": 15870
    },
    {
      "epoch": 0.13136451999834553,
      "grad_norm": 1166.7960205078125,
      "learning_rate": 9.975583321764638e-06,
      "loss": 144.3113,
      "step": 15880
    },
    {
      "epoch": 0.13144724324771476,
      "grad_norm": 1041.4771728515625,
      "learning_rate": 9.975443912490073e-06,
      "loss": 149.5042,
      "step": 15890
    },
    {
      "epoch": 0.131529966497084,
      "grad_norm": 2316.5087890625,
      "learning_rate": 9.975304107342268e-06,
      "loss": 179.2303,
      "step": 15900
    },
    {
      "epoch": 0.13161268974645324,
      "grad_norm": 692.1578369140625,
      "learning_rate": 9.97516390633235e-06,
      "loss": 133.4318,
      "step": 15910
    },
    {
      "epoch": 0.13169541299582246,
      "grad_norm": 1125.5006103515625,
      "learning_rate": 9.975023309471473e-06,
      "loss": 156.1001,
      "step": 15920
    },
    {
      "epoch": 0.13177813624519172,
      "grad_norm": 1064.159423828125,
      "learning_rate": 9.974882316770823e-06,
      "loss": 147.876,
      "step": 15930
    },
    {
      "epoch": 0.13186085949456094,
      "grad_norm": 1456.9761962890625,
      "learning_rate": 9.974740928241617e-06,
      "loss": 146.098,
      "step": 15940
    },
    {
      "epoch": 0.13194358274393017,
      "grad_norm": 1191.022705078125,
      "learning_rate": 9.974599143895107e-06,
      "loss": 139.6693,
      "step": 15950
    },
    {
      "epoch": 0.13202630599329943,
      "grad_norm": 2010.4088134765625,
      "learning_rate": 9.974456963742573e-06,
      "loss": 152.4677,
      "step": 15960
    },
    {
      "epoch": 0.13210902924266865,
      "grad_norm": 1077.85205078125,
      "learning_rate": 9.97431438779533e-06,
      "loss": 199.6097,
      "step": 15970
    },
    {
      "epoch": 0.13219175249203788,
      "grad_norm": 975.5093994140625,
      "learning_rate": 9.974171416064719e-06,
      "loss": 110.029,
      "step": 15980
    },
    {
      "epoch": 0.13227447574140713,
      "grad_norm": 1180.7437744140625,
      "learning_rate": 9.974028048562118e-06,
      "loss": 136.7102,
      "step": 15990
    },
    {
      "epoch": 0.13235719899077636,
      "grad_norm": 1241.7110595703125,
      "learning_rate": 9.973884285298932e-06,
      "loss": 154.7749,
      "step": 16000
    },
    {
      "epoch": 0.13243992224014559,
      "grad_norm": 1181.715576171875,
      "learning_rate": 9.9737401262866e-06,
      "loss": 163.9049,
      "step": 16010
    },
    {
      "epoch": 0.13252264548951484,
      "grad_norm": 1441.7060546875,
      "learning_rate": 9.973595571536593e-06,
      "loss": 131.6654,
      "step": 16020
    },
    {
      "epoch": 0.13260536873888407,
      "grad_norm": 1810.145751953125,
      "learning_rate": 9.973450621060412e-06,
      "loss": 155.4361,
      "step": 16030
    },
    {
      "epoch": 0.1326880919882533,
      "grad_norm": 1024.084716796875,
      "learning_rate": 9.97330527486959e-06,
      "loss": 130.6234,
      "step": 16040
    },
    {
      "epoch": 0.13277081523762255,
      "grad_norm": 1294.561279296875,
      "learning_rate": 9.973159532975691e-06,
      "loss": 122.6079,
      "step": 16050
    },
    {
      "epoch": 0.13285353848699177,
      "grad_norm": 1282.573486328125,
      "learning_rate": 9.973013395390314e-06,
      "loss": 173.6021,
      "step": 16060
    },
    {
      "epoch": 0.132936261736361,
      "grad_norm": 1436.6795654296875,
      "learning_rate": 9.972866862125083e-06,
      "loss": 201.6667,
      "step": 16070
    },
    {
      "epoch": 0.13301898498573023,
      "grad_norm": 880.5997924804688,
      "learning_rate": 9.972719933191657e-06,
      "loss": 121.1312,
      "step": 16080
    },
    {
      "epoch": 0.13310170823509948,
      "grad_norm": 720.5911254882812,
      "learning_rate": 9.97257260860173e-06,
      "loss": 117.1484,
      "step": 16090
    },
    {
      "epoch": 0.1331844314844687,
      "grad_norm": 1505.3927001953125,
      "learning_rate": 9.972424888367019e-06,
      "loss": 146.7309,
      "step": 16100
    },
    {
      "epoch": 0.13326715473383793,
      "grad_norm": 958.6402587890625,
      "learning_rate": 9.972276772499281e-06,
      "loss": 156.9766,
      "step": 16110
    },
    {
      "epoch": 0.1333498779832072,
      "grad_norm": 877.50244140625,
      "learning_rate": 9.9721282610103e-06,
      "loss": 191.0899,
      "step": 16120
    },
    {
      "epoch": 0.13343260123257641,
      "grad_norm": 1021.2138671875,
      "learning_rate": 9.971979353911891e-06,
      "loss": 133.9165,
      "step": 16130
    },
    {
      "epoch": 0.13351532448194564,
      "grad_norm": 847.0870971679688,
      "learning_rate": 9.971830051215905e-06,
      "loss": 101.3374,
      "step": 16140
    },
    {
      "epoch": 0.1335980477313149,
      "grad_norm": 2785.597412109375,
      "learning_rate": 9.97168035293422e-06,
      "loss": 267.7292,
      "step": 16150
    },
    {
      "epoch": 0.13368077098068412,
      "grad_norm": 801.3421020507812,
      "learning_rate": 9.971530259078743e-06,
      "loss": 111.4734,
      "step": 16160
    },
    {
      "epoch": 0.13376349423005335,
      "grad_norm": 768.2542114257812,
      "learning_rate": 9.971379769661422e-06,
      "loss": 149.4196,
      "step": 16170
    },
    {
      "epoch": 0.1338462174794226,
      "grad_norm": 893.0291748046875,
      "learning_rate": 9.971228884694228e-06,
      "loss": 122.37,
      "step": 16180
    },
    {
      "epoch": 0.13392894072879183,
      "grad_norm": 1295.072509765625,
      "learning_rate": 9.971077604189166e-06,
      "loss": 156.3286,
      "step": 16190
    },
    {
      "epoch": 0.13401166397816106,
      "grad_norm": 998.2085571289062,
      "learning_rate": 9.970925928158275e-06,
      "loss": 122.403,
      "step": 16200
    },
    {
      "epoch": 0.1340943872275303,
      "grad_norm": 701.3370361328125,
      "learning_rate": 9.970773856613617e-06,
      "loss": 140.6802,
      "step": 16210
    },
    {
      "epoch": 0.13417711047689954,
      "grad_norm": 971.6983032226562,
      "learning_rate": 9.970621389567301e-06,
      "loss": 178.1052,
      "step": 16220
    },
    {
      "epoch": 0.13425983372626876,
      "grad_norm": 2665.119384765625,
      "learning_rate": 9.97046852703145e-06,
      "loss": 138.6044,
      "step": 16230
    },
    {
      "epoch": 0.13434255697563802,
      "grad_norm": 2127.31884765625,
      "learning_rate": 9.970315269018231e-06,
      "loss": 157.2493,
      "step": 16240
    },
    {
      "epoch": 0.13442528022500724,
      "grad_norm": 1778.2391357421875,
      "learning_rate": 9.970161615539837e-06,
      "loss": 134.0471,
      "step": 16250
    },
    {
      "epoch": 0.13450800347437647,
      "grad_norm": 993.4716796875,
      "learning_rate": 9.970007566608492e-06,
      "loss": 146.2506,
      "step": 16260
    },
    {
      "epoch": 0.1345907267237457,
      "grad_norm": 798.2664184570312,
      "learning_rate": 9.969853122236455e-06,
      "loss": 114.1296,
      "step": 16270
    },
    {
      "epoch": 0.13467344997311495,
      "grad_norm": 703.0869750976562,
      "learning_rate": 9.969698282436013e-06,
      "loss": 120.5299,
      "step": 16280
    },
    {
      "epoch": 0.13475617322248418,
      "grad_norm": 1201.6317138671875,
      "learning_rate": 9.969543047219487e-06,
      "loss": 125.8007,
      "step": 16290
    },
    {
      "epoch": 0.1348388964718534,
      "grad_norm": 1785.0177001953125,
      "learning_rate": 9.969387416599227e-06,
      "loss": 144.5029,
      "step": 16300
    },
    {
      "epoch": 0.13492161972122266,
      "grad_norm": 1228.9619140625,
      "learning_rate": 9.969231390587618e-06,
      "loss": 164.9693,
      "step": 16310
    },
    {
      "epoch": 0.13500434297059188,
      "grad_norm": 864.3604736328125,
      "learning_rate": 9.969074969197072e-06,
      "loss": 168.7043,
      "step": 16320
    },
    {
      "epoch": 0.1350870662199611,
      "grad_norm": 1214.023681640625,
      "learning_rate": 9.968918152440036e-06,
      "loss": 172.751,
      "step": 16330
    },
    {
      "epoch": 0.13516978946933036,
      "grad_norm": 928.501220703125,
      "learning_rate": 9.968760940328987e-06,
      "loss": 131.5311,
      "step": 16340
    },
    {
      "epoch": 0.1352525127186996,
      "grad_norm": 510.1147155761719,
      "learning_rate": 9.968603332876435e-06,
      "loss": 171.1721,
      "step": 16350
    },
    {
      "epoch": 0.13533523596806882,
      "grad_norm": 1110.3807373046875,
      "learning_rate": 9.968445330094915e-06,
      "loss": 169.255,
      "step": 16360
    },
    {
      "epoch": 0.13541795921743807,
      "grad_norm": 1672.8614501953125,
      "learning_rate": 9.968286931997004e-06,
      "loss": 112.5926,
      "step": 16370
    },
    {
      "epoch": 0.1355006824668073,
      "grad_norm": 1014.0128784179688,
      "learning_rate": 9.968128138595304e-06,
      "loss": 100.9882,
      "step": 16380
    },
    {
      "epoch": 0.13558340571617652,
      "grad_norm": 1446.147216796875,
      "learning_rate": 9.967968949902448e-06,
      "loss": 185.0402,
      "step": 16390
    },
    {
      "epoch": 0.13566612896554578,
      "grad_norm": 753.0343627929688,
      "learning_rate": 9.967809365931102e-06,
      "loss": 148.759,
      "step": 16400
    },
    {
      "epoch": 0.135748852214915,
      "grad_norm": 909.8871459960938,
      "learning_rate": 9.967649386693964e-06,
      "loss": 123.6662,
      "step": 16410
    },
    {
      "epoch": 0.13583157546428423,
      "grad_norm": 1223.7244873046875,
      "learning_rate": 9.967489012203765e-06,
      "loss": 132.6178,
      "step": 16420
    },
    {
      "epoch": 0.13591429871365346,
      "grad_norm": 1106.0858154296875,
      "learning_rate": 9.967328242473261e-06,
      "loss": 146.9553,
      "step": 16430
    },
    {
      "epoch": 0.1359970219630227,
      "grad_norm": 1789.8829345703125,
      "learning_rate": 9.967167077515246e-06,
      "loss": 133.0784,
      "step": 16440
    },
    {
      "epoch": 0.13607974521239194,
      "grad_norm": 741.1414184570312,
      "learning_rate": 9.967005517342544e-06,
      "loss": 143.1583,
      "step": 16450
    },
    {
      "epoch": 0.13616246846176117,
      "grad_norm": 1324.021240234375,
      "learning_rate": 9.966843561968005e-06,
      "loss": 108.1861,
      "step": 16460
    },
    {
      "epoch": 0.13624519171113042,
      "grad_norm": 866.0011596679688,
      "learning_rate": 9.966681211404521e-06,
      "loss": 138.6324,
      "step": 16470
    },
    {
      "epoch": 0.13632791496049965,
      "grad_norm": 520.3377685546875,
      "learning_rate": 9.966518465665007e-06,
      "loss": 113.3134,
      "step": 16480
    },
    {
      "epoch": 0.13641063820986887,
      "grad_norm": 883.1153564453125,
      "learning_rate": 9.966355324762412e-06,
      "loss": 163.313,
      "step": 16490
    },
    {
      "epoch": 0.13649336145923813,
      "grad_norm": 1007.1843872070312,
      "learning_rate": 9.966191788709716e-06,
      "loss": 140.2184,
      "step": 16500
    },
    {
      "epoch": 0.13657608470860735,
      "grad_norm": 1669.2816162109375,
      "learning_rate": 9.966027857519931e-06,
      "loss": 188.2176,
      "step": 16510
    },
    {
      "epoch": 0.13665880795797658,
      "grad_norm": 772.6116943359375,
      "learning_rate": 9.9658635312061e-06,
      "loss": 163.7544,
      "step": 16520
    },
    {
      "epoch": 0.13674153120734583,
      "grad_norm": 706.4850463867188,
      "learning_rate": 9.965698809781298e-06,
      "loss": 121.3989,
      "step": 16530
    },
    {
      "epoch": 0.13682425445671506,
      "grad_norm": 766.0828247070312,
      "learning_rate": 9.965533693258632e-06,
      "loss": 213.4713,
      "step": 16540
    },
    {
      "epoch": 0.1369069777060843,
      "grad_norm": 957.917724609375,
      "learning_rate": 9.965368181651239e-06,
      "loss": 183.1273,
      "step": 16550
    },
    {
      "epoch": 0.13698970095545354,
      "grad_norm": 696.8062744140625,
      "learning_rate": 9.965202274972288e-06,
      "loss": 112.6891,
      "step": 16560
    },
    {
      "epoch": 0.13707242420482277,
      "grad_norm": 902.3621215820312,
      "learning_rate": 9.965035973234977e-06,
      "loss": 113.6838,
      "step": 16570
    },
    {
      "epoch": 0.137155147454192,
      "grad_norm": 1020.390625,
      "learning_rate": 9.964869276452542e-06,
      "loss": 106.0109,
      "step": 16580
    },
    {
      "epoch": 0.13723787070356125,
      "grad_norm": 1181.8326416015625,
      "learning_rate": 9.964702184638244e-06,
      "loss": 139.7021,
      "step": 16590
    },
    {
      "epoch": 0.13732059395293048,
      "grad_norm": 629.9285278320312,
      "learning_rate": 9.964534697805377e-06,
      "loss": 193.1732,
      "step": 16600
    },
    {
      "epoch": 0.1374033172022997,
      "grad_norm": 1531.7962646484375,
      "learning_rate": 9.96436681596727e-06,
      "loss": 154.7776,
      "step": 16610
    },
    {
      "epoch": 0.13748604045166893,
      "grad_norm": 1220.1796875,
      "learning_rate": 9.964198539137277e-06,
      "loss": 191.2195,
      "step": 16620
    },
    {
      "epoch": 0.13756876370103818,
      "grad_norm": 0.0,
      "learning_rate": 9.964029867328791e-06,
      "loss": 112.8693,
      "step": 16630
    },
    {
      "epoch": 0.1376514869504074,
      "grad_norm": 1105.817626953125,
      "learning_rate": 9.963860800555228e-06,
      "loss": 103.0777,
      "step": 16640
    },
    {
      "epoch": 0.13773421019977664,
      "grad_norm": 472.4584655761719,
      "learning_rate": 9.963691338830045e-06,
      "loss": 123.1952,
      "step": 16650
    },
    {
      "epoch": 0.1378169334491459,
      "grad_norm": 990.940673828125,
      "learning_rate": 9.963521482166718e-06,
      "loss": 136.4567,
      "step": 16660
    },
    {
      "epoch": 0.13789965669851512,
      "grad_norm": 1503.9461669921875,
      "learning_rate": 9.96335123057877e-06,
      "loss": 136.2858,
      "step": 16670
    },
    {
      "epoch": 0.13798237994788434,
      "grad_norm": 1348.58740234375,
      "learning_rate": 9.963180584079741e-06,
      "loss": 137.5341,
      "step": 16680
    },
    {
      "epoch": 0.1380651031972536,
      "grad_norm": 1100.0037841796875,
      "learning_rate": 9.963009542683214e-06,
      "loss": 199.9709,
      "step": 16690
    },
    {
      "epoch": 0.13814782644662282,
      "grad_norm": 718.8609619140625,
      "learning_rate": 9.962838106402791e-06,
      "loss": 184.6782,
      "step": 16700
    },
    {
      "epoch": 0.13823054969599205,
      "grad_norm": 865.5576782226562,
      "learning_rate": 9.962666275252117e-06,
      "loss": 104.1854,
      "step": 16710
    },
    {
      "epoch": 0.1383132729453613,
      "grad_norm": 1161.63525390625,
      "learning_rate": 9.962494049244866e-06,
      "loss": 169.3983,
      "step": 16720
    },
    {
      "epoch": 0.13839599619473053,
      "grad_norm": 589.0774536132812,
      "learning_rate": 9.962321428394735e-06,
      "loss": 165.776,
      "step": 16730
    },
    {
      "epoch": 0.13847871944409976,
      "grad_norm": 2693.160888671875,
      "learning_rate": 9.962148412715464e-06,
      "loss": 154.1448,
      "step": 16740
    },
    {
      "epoch": 0.138561442693469,
      "grad_norm": 1310.2269287109375,
      "learning_rate": 9.961975002220816e-06,
      "loss": 166.3599,
      "step": 16750
    },
    {
      "epoch": 0.13864416594283824,
      "grad_norm": 1167.153076171875,
      "learning_rate": 9.96180119692459e-06,
      "loss": 171.0495,
      "step": 16760
    },
    {
      "epoch": 0.13872688919220746,
      "grad_norm": 1377.29833984375,
      "learning_rate": 9.961626996840613e-06,
      "loss": 102.7167,
      "step": 16770
    },
    {
      "epoch": 0.13880961244157672,
      "grad_norm": 977.5831909179688,
      "learning_rate": 9.961452401982748e-06,
      "loss": 136.4004,
      "step": 16780
    },
    {
      "epoch": 0.13889233569094595,
      "grad_norm": 1010.1982421875,
      "learning_rate": 9.961277412364884e-06,
      "loss": 146.971,
      "step": 16790
    },
    {
      "epoch": 0.13897505894031517,
      "grad_norm": 814.7576293945312,
      "learning_rate": 9.961102028000948e-06,
      "loss": 213.2676,
      "step": 16800
    },
    {
      "epoch": 0.1390577821896844,
      "grad_norm": 881.7014770507812,
      "learning_rate": 9.96092624890489e-06,
      "loss": 91.0271,
      "step": 16810
    },
    {
      "epoch": 0.13914050543905365,
      "grad_norm": 4899.205078125,
      "learning_rate": 9.960750075090698e-06,
      "loss": 166.8467,
      "step": 16820
    },
    {
      "epoch": 0.13922322868842288,
      "grad_norm": 1270.030029296875,
      "learning_rate": 9.960573506572391e-06,
      "loss": 186.535,
      "step": 16830
    },
    {
      "epoch": 0.1393059519377921,
      "grad_norm": 1338.3089599609375,
      "learning_rate": 9.960396543364013e-06,
      "loss": 192.4324,
      "step": 16840
    },
    {
      "epoch": 0.13938867518716136,
      "grad_norm": 1512.3917236328125,
      "learning_rate": 9.96021918547965e-06,
      "loss": 124.9194,
      "step": 16850
    },
    {
      "epoch": 0.13947139843653059,
      "grad_norm": 1637.7535400390625,
      "learning_rate": 9.96004143293341e-06,
      "loss": 131.2566,
      "step": 16860
    },
    {
      "epoch": 0.1395541216858998,
      "grad_norm": 1564.211669921875,
      "learning_rate": 9.959863285739436e-06,
      "loss": 124.8255,
      "step": 16870
    },
    {
      "epoch": 0.13963684493526907,
      "grad_norm": 720.8834228515625,
      "learning_rate": 9.959684743911904e-06,
      "loss": 140.7759,
      "step": 16880
    },
    {
      "epoch": 0.1397195681846383,
      "grad_norm": 796.6300659179688,
      "learning_rate": 9.959505807465018e-06,
      "loss": 120.1176,
      "step": 16890
    },
    {
      "epoch": 0.13980229143400752,
      "grad_norm": 1232.4276123046875,
      "learning_rate": 9.959326476413016e-06,
      "loss": 130.2664,
      "step": 16900
    },
    {
      "epoch": 0.13988501468337677,
      "grad_norm": 457.3919677734375,
      "learning_rate": 9.959146750770167e-06,
      "loss": 124.8512,
      "step": 16910
    },
    {
      "epoch": 0.139967737932746,
      "grad_norm": 708.2092895507812,
      "learning_rate": 9.95896663055077e-06,
      "loss": 120.5444,
      "step": 16920
    },
    {
      "epoch": 0.14005046118211523,
      "grad_norm": 995.7003784179688,
      "learning_rate": 9.958786115769157e-06,
      "loss": 114.9213,
      "step": 16930
    },
    {
      "epoch": 0.14013318443148448,
      "grad_norm": 1515.4827880859375,
      "learning_rate": 9.958605206439692e-06,
      "loss": 146.7894,
      "step": 16940
    },
    {
      "epoch": 0.1402159076808537,
      "grad_norm": 814.6317138671875,
      "learning_rate": 9.958423902576764e-06,
      "loss": 99.1024,
      "step": 16950
    },
    {
      "epoch": 0.14029863093022293,
      "grad_norm": 760.5602416992188,
      "learning_rate": 9.958242204194804e-06,
      "loss": 160.827,
      "step": 16960
    },
    {
      "epoch": 0.1403813541795922,
      "grad_norm": 738.33349609375,
      "learning_rate": 9.958060111308267e-06,
      "loss": 136.0457,
      "step": 16970
    },
    {
      "epoch": 0.14046407742896141,
      "grad_norm": 1149.28857421875,
      "learning_rate": 9.957877623931642e-06,
      "loss": 151.1577,
      "step": 16980
    },
    {
      "epoch": 0.14054680067833064,
      "grad_norm": 1362.2108154296875,
      "learning_rate": 9.95769474207945e-06,
      "loss": 173.5694,
      "step": 16990
    },
    {
      "epoch": 0.14062952392769987,
      "grad_norm": 1314.1846923828125,
      "learning_rate": 9.957511465766236e-06,
      "loss": 169.4035,
      "step": 17000
    },
    {
      "epoch": 0.14071224717706912,
      "grad_norm": 1065.3922119140625,
      "learning_rate": 9.957327795006589e-06,
      "loss": 169.1779,
      "step": 17010
    },
    {
      "epoch": 0.14079497042643835,
      "grad_norm": 1382.638427734375,
      "learning_rate": 9.95714372981512e-06,
      "loss": 145.6161,
      "step": 17020
    },
    {
      "epoch": 0.14087769367580757,
      "grad_norm": 1485.4481201171875,
      "learning_rate": 9.956959270206474e-06,
      "loss": 131.7884,
      "step": 17030
    },
    {
      "epoch": 0.14096041692517683,
      "grad_norm": 901.7747192382812,
      "learning_rate": 9.956774416195329e-06,
      "loss": 129.2612,
      "step": 17040
    },
    {
      "epoch": 0.14104314017454606,
      "grad_norm": 1346.950439453125,
      "learning_rate": 9.956589167796392e-06,
      "loss": 108.1172,
      "step": 17050
    },
    {
      "epoch": 0.14112586342391528,
      "grad_norm": 857.8418579101562,
      "learning_rate": 9.956403525024402e-06,
      "loss": 132.697,
      "step": 17060
    },
    {
      "epoch": 0.14120858667328454,
      "grad_norm": 1938.1868896484375,
      "learning_rate": 9.956217487894131e-06,
      "loss": 165.6452,
      "step": 17070
    },
    {
      "epoch": 0.14129130992265376,
      "grad_norm": 749.3518676757812,
      "learning_rate": 9.95603105642038e-06,
      "loss": 212.4321,
      "step": 17080
    },
    {
      "epoch": 0.141374033172023,
      "grad_norm": 709.408447265625,
      "learning_rate": 9.955844230617985e-06,
      "loss": 156.41,
      "step": 17090
    },
    {
      "epoch": 0.14145675642139224,
      "grad_norm": 1008.6261596679688,
      "learning_rate": 9.955657010501807e-06,
      "loss": 118.0272,
      "step": 17100
    },
    {
      "epoch": 0.14153947967076147,
      "grad_norm": 828.895751953125,
      "learning_rate": 9.955469396086743e-06,
      "loss": 138.8411,
      "step": 17110
    },
    {
      "epoch": 0.1416222029201307,
      "grad_norm": 1362.32421875,
      "learning_rate": 9.955281387387724e-06,
      "loss": 145.7589,
      "step": 17120
    },
    {
      "epoch": 0.14170492616949995,
      "grad_norm": 1597.079345703125,
      "learning_rate": 9.955092984419705e-06,
      "loss": 170.475,
      "step": 17130
    },
    {
      "epoch": 0.14178764941886918,
      "grad_norm": 1059.4306640625,
      "learning_rate": 9.954904187197679e-06,
      "loss": 158.0434,
      "step": 17140
    },
    {
      "epoch": 0.1418703726682384,
      "grad_norm": 694.0506591796875,
      "learning_rate": 9.954714995736667e-06,
      "loss": 142.6755,
      "step": 17150
    },
    {
      "epoch": 0.14195309591760763,
      "grad_norm": 1392.7862548828125,
      "learning_rate": 9.95452541005172e-06,
      "loss": 192.9698,
      "step": 17160
    },
    {
      "epoch": 0.14203581916697688,
      "grad_norm": 1239.712646484375,
      "learning_rate": 9.954335430157926e-06,
      "loss": 126.2119,
      "step": 17170
    },
    {
      "epoch": 0.1421185424163461,
      "grad_norm": 949.230712890625,
      "learning_rate": 9.9541450560704e-06,
      "loss": 76.8772,
      "step": 17180
    },
    {
      "epoch": 0.14220126566571534,
      "grad_norm": 1190.4364013671875,
      "learning_rate": 9.953954287804286e-06,
      "loss": 156.7768,
      "step": 17190
    },
    {
      "epoch": 0.1422839889150846,
      "grad_norm": 1422.4742431640625,
      "learning_rate": 9.953763125374767e-06,
      "loss": 107.7513,
      "step": 17200
    },
    {
      "epoch": 0.14236671216445382,
      "grad_norm": 1076.408935546875,
      "learning_rate": 9.953571568797049e-06,
      "loss": 136.0641,
      "step": 17210
    },
    {
      "epoch": 0.14244943541382304,
      "grad_norm": 930.828125,
      "learning_rate": 9.953379618086377e-06,
      "loss": 143.9599,
      "step": 17220
    },
    {
      "epoch": 0.1425321586631923,
      "grad_norm": 1367.8873291015625,
      "learning_rate": 9.95318727325802e-06,
      "loss": 128.7768,
      "step": 17230
    },
    {
      "epoch": 0.14261488191256153,
      "grad_norm": 1150.171875,
      "learning_rate": 9.952994534327283e-06,
      "loss": 124.427,
      "step": 17240
    },
    {
      "epoch": 0.14269760516193075,
      "grad_norm": 821.237548828125,
      "learning_rate": 9.952801401309504e-06,
      "loss": 137.096,
      "step": 17250
    },
    {
      "epoch": 0.1427803284113,
      "grad_norm": 1357.8616943359375,
      "learning_rate": 9.952607874220048e-06,
      "loss": 201.047,
      "step": 17260
    },
    {
      "epoch": 0.14286305166066923,
      "grad_norm": 1452.91650390625,
      "learning_rate": 9.952413953074312e-06,
      "loss": 199.8793,
      "step": 17270
    },
    {
      "epoch": 0.14294577491003846,
      "grad_norm": 965.8828125,
      "learning_rate": 9.952219637887725e-06,
      "loss": 129.7407,
      "step": 17280
    },
    {
      "epoch": 0.1430284981594077,
      "grad_norm": 1721.4344482421875,
      "learning_rate": 9.952024928675752e-06,
      "loss": 177.8543,
      "step": 17290
    },
    {
      "epoch": 0.14311122140877694,
      "grad_norm": 3541.317626953125,
      "learning_rate": 9.951829825453881e-06,
      "loss": 167.7698,
      "step": 17300
    },
    {
      "epoch": 0.14319394465814617,
      "grad_norm": 2036.2423095703125,
      "learning_rate": 9.951634328237635e-06,
      "loss": 141.8449,
      "step": 17310
    },
    {
      "epoch": 0.14327666790751542,
      "grad_norm": 880.5416870117188,
      "learning_rate": 9.951438437042572e-06,
      "loss": 198.8033,
      "step": 17320
    },
    {
      "epoch": 0.14335939115688465,
      "grad_norm": 807.236572265625,
      "learning_rate": 9.951242151884275e-06,
      "loss": 112.0078,
      "step": 17330
    },
    {
      "epoch": 0.14344211440625387,
      "grad_norm": 1530.7301025390625,
      "learning_rate": 9.951045472778365e-06,
      "loss": 133.3953,
      "step": 17340
    },
    {
      "epoch": 0.1435248376556231,
      "grad_norm": 1775.3485107421875,
      "learning_rate": 9.950848399740488e-06,
      "loss": 132.5112,
      "step": 17350
    },
    {
      "epoch": 0.14360756090499235,
      "grad_norm": 1216.1314697265625,
      "learning_rate": 9.950650932786325e-06,
      "loss": 150.7454,
      "step": 17360
    },
    {
      "epoch": 0.14369028415436158,
      "grad_norm": 756.1212158203125,
      "learning_rate": 9.95045307193159e-06,
      "loss": 114.4585,
      "step": 17370
    },
    {
      "epoch": 0.1437730074037308,
      "grad_norm": 987.248779296875,
      "learning_rate": 9.95025481719202e-06,
      "loss": 140.8504,
      "step": 17380
    },
    {
      "epoch": 0.14385573065310006,
      "grad_norm": 1126.249267578125,
      "learning_rate": 9.950056168583395e-06,
      "loss": 225.9696,
      "step": 17390
    },
    {
      "epoch": 0.1439384539024693,
      "grad_norm": 706.3463745117188,
      "learning_rate": 9.949857126121519e-06,
      "loss": 113.696,
      "step": 17400
    },
    {
      "epoch": 0.14402117715183851,
      "grad_norm": 892.3402099609375,
      "learning_rate": 9.949657689822226e-06,
      "loss": 162.9231,
      "step": 17410
    },
    {
      "epoch": 0.14410390040120777,
      "grad_norm": 856.6466674804688,
      "learning_rate": 9.949457859701388e-06,
      "loss": 99.4635,
      "step": 17420
    },
    {
      "epoch": 0.144186623650577,
      "grad_norm": 775.4996948242188,
      "learning_rate": 9.949257635774903e-06,
      "loss": 152.7363,
      "step": 17430
    },
    {
      "epoch": 0.14426934689994622,
      "grad_norm": 842.1768798828125,
      "learning_rate": 9.9490570180587e-06,
      "loss": 85.8346,
      "step": 17440
    },
    {
      "epoch": 0.14435207014931548,
      "grad_norm": 1798.95849609375,
      "learning_rate": 9.948856006568746e-06,
      "loss": 197.5757,
      "step": 17450
    },
    {
      "epoch": 0.1444347933986847,
      "grad_norm": 1381.5155029296875,
      "learning_rate": 9.94865460132103e-06,
      "loss": 150.2531,
      "step": 17460
    },
    {
      "epoch": 0.14451751664805393,
      "grad_norm": 997.7630004882812,
      "learning_rate": 9.948452802331578e-06,
      "loss": 133.1603,
      "step": 17470
    },
    {
      "epoch": 0.14460023989742318,
      "grad_norm": 1275.1690673828125,
      "learning_rate": 9.948250609616449e-06,
      "loss": 168.5733,
      "step": 17480
    },
    {
      "epoch": 0.1446829631467924,
      "grad_norm": 1112.8721923828125,
      "learning_rate": 9.948048023191728e-06,
      "loss": 182.301,
      "step": 17490
    },
    {
      "epoch": 0.14476568639616164,
      "grad_norm": 950.4414672851562,
      "learning_rate": 9.947845043073533e-06,
      "loss": 149.5477,
      "step": 17500
    },
    {
      "epoch": 0.1448484096455309,
      "grad_norm": 1122.95751953125,
      "learning_rate": 9.947641669278016e-06,
      "loss": 123.1119,
      "step": 17510
    },
    {
      "epoch": 0.14493113289490012,
      "grad_norm": 1148.9334716796875,
      "learning_rate": 9.947437901821358e-06,
      "loss": 128.3063,
      "step": 17520
    },
    {
      "epoch": 0.14501385614426934,
      "grad_norm": 1392.179443359375,
      "learning_rate": 9.947233740719772e-06,
      "loss": 139.3278,
      "step": 17530
    },
    {
      "epoch": 0.14509657939363857,
      "grad_norm": 604.5231323242188,
      "learning_rate": 9.947029185989501e-06,
      "loss": 163.2896,
      "step": 17540
    },
    {
      "epoch": 0.14517930264300782,
      "grad_norm": 1102.948486328125,
      "learning_rate": 9.946824237646823e-06,
      "loss": 153.8839,
      "step": 17550
    },
    {
      "epoch": 0.14526202589237705,
      "grad_norm": 2167.79638671875,
      "learning_rate": 9.946618895708043e-06,
      "loss": 172.0367,
      "step": 17560
    },
    {
      "epoch": 0.14534474914174628,
      "grad_norm": 1476.7362060546875,
      "learning_rate": 9.946413160189498e-06,
      "loss": 138.3295,
      "step": 17570
    },
    {
      "epoch": 0.14542747239111553,
      "grad_norm": 882.1810913085938,
      "learning_rate": 9.946207031107562e-06,
      "loss": 186.2194,
      "step": 17580
    },
    {
      "epoch": 0.14551019564048476,
      "grad_norm": 2111.673095703125,
      "learning_rate": 9.94600050847863e-06,
      "loss": 170.1872,
      "step": 17590
    },
    {
      "epoch": 0.14559291888985398,
      "grad_norm": 1099.032958984375,
      "learning_rate": 9.945793592319137e-06,
      "loss": 128.6498,
      "step": 17600
    },
    {
      "epoch": 0.14567564213922324,
      "grad_norm": 1059.4005126953125,
      "learning_rate": 9.945586282645545e-06,
      "loss": 134.5357,
      "step": 17610
    },
    {
      "epoch": 0.14575836538859246,
      "grad_norm": 1566.564208984375,
      "learning_rate": 9.945378579474351e-06,
      "loss": 164.359,
      "step": 17620
    },
    {
      "epoch": 0.1458410886379617,
      "grad_norm": 782.61279296875,
      "learning_rate": 9.945170482822079e-06,
      "loss": 106.899,
      "step": 17630
    },
    {
      "epoch": 0.14592381188733095,
      "grad_norm": 1026.7816162109375,
      "learning_rate": 9.944961992705288e-06,
      "loss": 142.0462,
      "step": 17640
    },
    {
      "epoch": 0.14600653513670017,
      "grad_norm": 817.039306640625,
      "learning_rate": 9.944753109140564e-06,
      "loss": 166.4367,
      "step": 17650
    },
    {
      "epoch": 0.1460892583860694,
      "grad_norm": 856.3842163085938,
      "learning_rate": 9.94454383214453e-06,
      "loss": 131.3289,
      "step": 17660
    },
    {
      "epoch": 0.14617198163543865,
      "grad_norm": 1656.999755859375,
      "learning_rate": 9.944334161733835e-06,
      "loss": 129.1978,
      "step": 17670
    },
    {
      "epoch": 0.14625470488480788,
      "grad_norm": 1338.8382568359375,
      "learning_rate": 9.944124097925161e-06,
      "loss": 184.4288,
      "step": 17680
    },
    {
      "epoch": 0.1463374281341771,
      "grad_norm": 1011.0686645507812,
      "learning_rate": 9.943913640735224e-06,
      "loss": 127.4451,
      "step": 17690
    },
    {
      "epoch": 0.14642015138354633,
      "grad_norm": 923.1884765625,
      "learning_rate": 9.94370279018077e-06,
      "loss": 120.5529,
      "step": 17700
    },
    {
      "epoch": 0.1465028746329156,
      "grad_norm": 1001.093505859375,
      "learning_rate": 9.94349154627857e-06,
      "loss": 129.3988,
      "step": 17710
    },
    {
      "epoch": 0.1465855978822848,
      "grad_norm": 1354.6356201171875,
      "learning_rate": 9.943279909045438e-06,
      "loss": 122.9835,
      "step": 17720
    },
    {
      "epoch": 0.14666832113165404,
      "grad_norm": 1260.7392578125,
      "learning_rate": 9.94306787849821e-06,
      "loss": 101.6319,
      "step": 17730
    },
    {
      "epoch": 0.1467510443810233,
      "grad_norm": 1424.63330078125,
      "learning_rate": 9.942855454653755e-06,
      "loss": 179.1118,
      "step": 17740
    },
    {
      "epoch": 0.14683376763039252,
      "grad_norm": 1053.8809814453125,
      "learning_rate": 9.942642637528977e-06,
      "loss": 167.5939,
      "step": 17750
    },
    {
      "epoch": 0.14691649087976175,
      "grad_norm": 936.3515014648438,
      "learning_rate": 9.942429427140807e-06,
      "loss": 154.7948,
      "step": 17760
    },
    {
      "epoch": 0.146999214129131,
      "grad_norm": 876.3916015625,
      "learning_rate": 9.942215823506211e-06,
      "loss": 114.5722,
      "step": 17770
    },
    {
      "epoch": 0.14708193737850023,
      "grad_norm": 1172.0423583984375,
      "learning_rate": 9.942001826642184e-06,
      "loss": 142.9646,
      "step": 17780
    },
    {
      "epoch": 0.14716466062786945,
      "grad_norm": 1635.97802734375,
      "learning_rate": 9.941787436565751e-06,
      "loss": 150.69,
      "step": 17790
    },
    {
      "epoch": 0.1472473838772387,
      "grad_norm": 728.3792724609375,
      "learning_rate": 9.941572653293974e-06,
      "loss": 97.5937,
      "step": 17800
    },
    {
      "epoch": 0.14733010712660793,
      "grad_norm": 935.0343627929688,
      "learning_rate": 9.941357476843938e-06,
      "loss": 135.0443,
      "step": 17810
    },
    {
      "epoch": 0.14741283037597716,
      "grad_norm": 583.3887329101562,
      "learning_rate": 9.941141907232766e-06,
      "loss": 134.4311,
      "step": 17820
    },
    {
      "epoch": 0.14749555362534642,
      "grad_norm": 1191.19677734375,
      "learning_rate": 9.940925944477608e-06,
      "loss": 129.727,
      "step": 17830
    },
    {
      "epoch": 0.14757827687471564,
      "grad_norm": 1111.1417236328125,
      "learning_rate": 9.940709588595649e-06,
      "loss": 171.4274,
      "step": 17840
    },
    {
      "epoch": 0.14766100012408487,
      "grad_norm": 2006.4134521484375,
      "learning_rate": 9.940492839604103e-06,
      "loss": 152.9817,
      "step": 17850
    },
    {
      "epoch": 0.14774372337345412,
      "grad_norm": 1163.596923828125,
      "learning_rate": 9.940275697520216e-06,
      "loss": 169.9584,
      "step": 17860
    },
    {
      "epoch": 0.14782644662282335,
      "grad_norm": 1189.015869140625,
      "learning_rate": 9.940058162361264e-06,
      "loss": 152.1794,
      "step": 17870
    },
    {
      "epoch": 0.14790916987219258,
      "grad_norm": 998.8855590820312,
      "learning_rate": 9.939840234144556e-06,
      "loss": 129.5204,
      "step": 17880
    },
    {
      "epoch": 0.1479918931215618,
      "grad_norm": 442.9149475097656,
      "learning_rate": 9.939621912887431e-06,
      "loss": 106.7805,
      "step": 17890
    },
    {
      "epoch": 0.14807461637093106,
      "grad_norm": 830.00927734375,
      "learning_rate": 9.93940319860726e-06,
      "loss": 139.6457,
      "step": 17900
    },
    {
      "epoch": 0.14815733962030028,
      "grad_norm": 1069.5220947265625,
      "learning_rate": 9.939184091321445e-06,
      "loss": 129.1493,
      "step": 17910
    },
    {
      "epoch": 0.1482400628696695,
      "grad_norm": 1180.868896484375,
      "learning_rate": 9.938964591047421e-06,
      "loss": 108.2578,
      "step": 17920
    },
    {
      "epoch": 0.14832278611903876,
      "grad_norm": 1095.6793212890625,
      "learning_rate": 9.938744697802651e-06,
      "loss": 145.4649,
      "step": 17930
    },
    {
      "epoch": 0.148405509368408,
      "grad_norm": 1292.62744140625,
      "learning_rate": 9.938524411604631e-06,
      "loss": 145.161,
      "step": 17940
    },
    {
      "epoch": 0.14848823261777722,
      "grad_norm": 1319.2213134765625,
      "learning_rate": 9.938303732470888e-06,
      "loss": 129.5037,
      "step": 17950
    },
    {
      "epoch": 0.14857095586714647,
      "grad_norm": 697.8318481445312,
      "learning_rate": 9.938082660418981e-06,
      "loss": 103.5571,
      "step": 17960
    },
    {
      "epoch": 0.1486536791165157,
      "grad_norm": 784.6300659179688,
      "learning_rate": 9.937861195466498e-06,
      "loss": 133.7046,
      "step": 17970
    },
    {
      "epoch": 0.14873640236588492,
      "grad_norm": 966.1806030273438,
      "learning_rate": 9.937639337631064e-06,
      "loss": 170.2544,
      "step": 17980
    },
    {
      "epoch": 0.14881912561525418,
      "grad_norm": 862.203857421875,
      "learning_rate": 9.937417086930328e-06,
      "loss": 129.5846,
      "step": 17990
    },
    {
      "epoch": 0.1489018488646234,
      "grad_norm": 3391.59716796875,
      "learning_rate": 9.937194443381972e-06,
      "loss": 195.0929,
      "step": 18000
    },
    {
      "epoch": 0.14898457211399263,
      "grad_norm": 737.4010009765625,
      "learning_rate": 9.936971407003714e-06,
      "loss": 110.9804,
      "step": 18010
    },
    {
      "epoch": 0.14906729536336188,
      "grad_norm": 1527.3822021484375,
      "learning_rate": 9.936747977813299e-06,
      "loss": 124.5241,
      "step": 18020
    },
    {
      "epoch": 0.1491500186127311,
      "grad_norm": 1383.10986328125,
      "learning_rate": 9.936524155828503e-06,
      "loss": 138.0007,
      "step": 18030
    },
    {
      "epoch": 0.14923274186210034,
      "grad_norm": 850.4631958007812,
      "learning_rate": 9.936299941067137e-06,
      "loss": 131.9197,
      "step": 18040
    },
    {
      "epoch": 0.1493154651114696,
      "grad_norm": 736.7586059570312,
      "learning_rate": 9.93607533354704e-06,
      "loss": 116.8003,
      "step": 18050
    },
    {
      "epoch": 0.14939818836083882,
      "grad_norm": 3558.953857421875,
      "learning_rate": 9.935850333286081e-06,
      "loss": 236.4352,
      "step": 18060
    },
    {
      "epoch": 0.14948091161020804,
      "grad_norm": 2170.6923828125,
      "learning_rate": 9.935624940302165e-06,
      "loss": 162.2385,
      "step": 18070
    },
    {
      "epoch": 0.14956363485957727,
      "grad_norm": 890.1776123046875,
      "learning_rate": 9.93539915461322e-06,
      "loss": 152.4946,
      "step": 18080
    },
    {
      "epoch": 0.14964635810894653,
      "grad_norm": 1055.08447265625,
      "learning_rate": 9.935172976237218e-06,
      "loss": 179.4581,
      "step": 18090
    },
    {
      "epoch": 0.14972908135831575,
      "grad_norm": 1069.735595703125,
      "learning_rate": 9.934946405192152e-06,
      "loss": 109.1896,
      "step": 18100
    },
    {
      "epoch": 0.14981180460768498,
      "grad_norm": 936.50048828125,
      "learning_rate": 9.934719441496048e-06,
      "loss": 170.2172,
      "step": 18110
    },
    {
      "epoch": 0.14989452785705423,
      "grad_norm": 1424.594970703125,
      "learning_rate": 9.934492085166965e-06,
      "loss": 120.5943,
      "step": 18120
    },
    {
      "epoch": 0.14997725110642346,
      "grad_norm": 1259.637939453125,
      "learning_rate": 9.934264336222992e-06,
      "loss": 141.8418,
      "step": 18130
    },
    {
      "epoch": 0.15005997435579269,
      "grad_norm": 1107.58447265625,
      "learning_rate": 9.934036194682253e-06,
      "loss": 132.8073,
      "step": 18140
    },
    {
      "epoch": 0.15014269760516194,
      "grad_norm": 1031.5169677734375,
      "learning_rate": 9.933807660562898e-06,
      "loss": 122.9906,
      "step": 18150
    },
    {
      "epoch": 0.15022542085453117,
      "grad_norm": 683.3692626953125,
      "learning_rate": 9.933578733883109e-06,
      "loss": 175.373,
      "step": 18160
    },
    {
      "epoch": 0.1503081441039004,
      "grad_norm": 841.3174438476562,
      "learning_rate": 9.933349414661103e-06,
      "loss": 143.8702,
      "step": 18170
    },
    {
      "epoch": 0.15039086735326965,
      "grad_norm": 1086.541015625,
      "learning_rate": 9.933119702915125e-06,
      "loss": 149.0898,
      "step": 18180
    },
    {
      "epoch": 0.15047359060263887,
      "grad_norm": 1380.8690185546875,
      "learning_rate": 9.932889598663452e-06,
      "loss": 142.0298,
      "step": 18190
    },
    {
      "epoch": 0.1505563138520081,
      "grad_norm": 800.4336547851562,
      "learning_rate": 9.932659101924393e-06,
      "loss": 169.3204,
      "step": 18200
    },
    {
      "epoch": 0.15063903710137735,
      "grad_norm": 883.3157348632812,
      "learning_rate": 9.932428212716287e-06,
      "loss": 183.8594,
      "step": 18210
    },
    {
      "epoch": 0.15072176035074658,
      "grad_norm": 922.6904907226562,
      "learning_rate": 9.932196931057505e-06,
      "loss": 157.8369,
      "step": 18220
    },
    {
      "epoch": 0.1508044836001158,
      "grad_norm": 1918.9375,
      "learning_rate": 9.931965256966449e-06,
      "loss": 143.9471,
      "step": 18230
    },
    {
      "epoch": 0.15088720684948506,
      "grad_norm": 1153.788818359375,
      "learning_rate": 9.931733190461552e-06,
      "loss": 167.4599,
      "step": 18240
    },
    {
      "epoch": 0.1509699300988543,
      "grad_norm": 1510.7779541015625,
      "learning_rate": 9.931500731561279e-06,
      "loss": 123.1982,
      "step": 18250
    },
    {
      "epoch": 0.15105265334822351,
      "grad_norm": 939.08447265625,
      "learning_rate": 9.931267880284124e-06,
      "loss": 128.6788,
      "step": 18260
    },
    {
      "epoch": 0.15113537659759274,
      "grad_norm": 521.10693359375,
      "learning_rate": 9.931034636648616e-06,
      "loss": 110.3548,
      "step": 18270
    },
    {
      "epoch": 0.151218099846962,
      "grad_norm": 1793.3514404296875,
      "learning_rate": 9.930801000673314e-06,
      "loss": 226.4601,
      "step": 18280
    },
    {
      "epoch": 0.15130082309633122,
      "grad_norm": 1056.244384765625,
      "learning_rate": 9.930566972376803e-06,
      "loss": 137.7991,
      "step": 18290
    },
    {
      "epoch": 0.15138354634570045,
      "grad_norm": 1053.8623046875,
      "learning_rate": 9.930332551777709e-06,
      "loss": 126.393,
      "step": 18300
    },
    {
      "epoch": 0.1514662695950697,
      "grad_norm": 466.3129577636719,
      "learning_rate": 9.930097738894679e-06,
      "loss": 142.9212,
      "step": 18310
    },
    {
      "epoch": 0.15154899284443893,
      "grad_norm": 1002.7549438476562,
      "learning_rate": 9.929862533746398e-06,
      "loss": 142.7721,
      "step": 18320
    },
    {
      "epoch": 0.15163171609380816,
      "grad_norm": 758.2431030273438,
      "learning_rate": 9.92962693635158e-06,
      "loss": 138.2474,
      "step": 18330
    },
    {
      "epoch": 0.1517144393431774,
      "grad_norm": 640.2601318359375,
      "learning_rate": 9.929390946728972e-06,
      "loss": 127.6863,
      "step": 18340
    },
    {
      "epoch": 0.15179716259254664,
      "grad_norm": 974.4703979492188,
      "learning_rate": 9.929154564897347e-06,
      "loss": 127.8559,
      "step": 18350
    },
    {
      "epoch": 0.15187988584191586,
      "grad_norm": 989.0883178710938,
      "learning_rate": 9.928917790875519e-06,
      "loss": 146.4885,
      "step": 18360
    },
    {
      "epoch": 0.15196260909128512,
      "grad_norm": 926.8894653320312,
      "learning_rate": 9.92868062468232e-06,
      "loss": 140.2718,
      "step": 18370
    },
    {
      "epoch": 0.15204533234065434,
      "grad_norm": 1093.3875732421875,
      "learning_rate": 9.928443066336624e-06,
      "loss": 156.4275,
      "step": 18380
    },
    {
      "epoch": 0.15212805559002357,
      "grad_norm": 1011.3756713867188,
      "learning_rate": 9.92820511585733e-06,
      "loss": 110.9518,
      "step": 18390
    },
    {
      "epoch": 0.15221077883939282,
      "grad_norm": 963.0042114257812,
      "learning_rate": 9.927966773263375e-06,
      "loss": 126.9806,
      "step": 18400
    },
    {
      "epoch": 0.15229350208876205,
      "grad_norm": 1338.420654296875,
      "learning_rate": 9.92772803857372e-06,
      "loss": 151.1979,
      "step": 18410
    },
    {
      "epoch": 0.15237622533813128,
      "grad_norm": 2231.09326171875,
      "learning_rate": 9.927488911807359e-06,
      "loss": 126.2641,
      "step": 18420
    },
    {
      "epoch": 0.1524589485875005,
      "grad_norm": 740.568603515625,
      "learning_rate": 9.927249392983319e-06,
      "loss": 161.8315,
      "step": 18430
    },
    {
      "epoch": 0.15254167183686976,
      "grad_norm": 1194.8526611328125,
      "learning_rate": 9.927009482120658e-06,
      "loss": 147.5258,
      "step": 18440
    },
    {
      "epoch": 0.15262439508623898,
      "grad_norm": 575.281005859375,
      "learning_rate": 9.926769179238467e-06,
      "loss": 123.2295,
      "step": 18450
    },
    {
      "epoch": 0.1527071183356082,
      "grad_norm": 1439.9266357421875,
      "learning_rate": 9.926528484355859e-06,
      "loss": 131.7167,
      "step": 18460
    },
    {
      "epoch": 0.15278984158497746,
      "grad_norm": 1190.1434326171875,
      "learning_rate": 9.926287397491992e-06,
      "loss": 147.6172,
      "step": 18470
    },
    {
      "epoch": 0.1528725648343467,
      "grad_norm": 1017.2939453125,
      "learning_rate": 9.926045918666045e-06,
      "loss": 144.2414,
      "step": 18480
    },
    {
      "epoch": 0.15295528808371592,
      "grad_norm": 1303.34814453125,
      "learning_rate": 9.925804047897231e-06,
      "loss": 202.542,
      "step": 18490
    },
    {
      "epoch": 0.15303801133308517,
      "grad_norm": 1302.6015625,
      "learning_rate": 9.925561785204797e-06,
      "loss": 150.7994,
      "step": 18500
    },
    {
      "epoch": 0.1531207345824544,
      "grad_norm": 1482.454345703125,
      "learning_rate": 9.925319130608015e-06,
      "loss": 160.8186,
      "step": 18510
    },
    {
      "epoch": 0.15320345783182363,
      "grad_norm": 1089.9215087890625,
      "learning_rate": 9.925076084126194e-06,
      "loss": 140.3311,
      "step": 18520
    },
    {
      "epoch": 0.15328618108119288,
      "grad_norm": 2586.4873046875,
      "learning_rate": 9.924832645778674e-06,
      "loss": 105.1053,
      "step": 18530
    },
    {
      "epoch": 0.1533689043305621,
      "grad_norm": 986.44775390625,
      "learning_rate": 9.924588815584822e-06,
      "loss": 146.6998,
      "step": 18540
    },
    {
      "epoch": 0.15345162757993133,
      "grad_norm": 1076.886474609375,
      "learning_rate": 9.924344593564038e-06,
      "loss": 167.1004,
      "step": 18550
    },
    {
      "epoch": 0.1535343508293006,
      "grad_norm": 576.464599609375,
      "learning_rate": 9.924099979735754e-06,
      "loss": 109.8678,
      "step": 18560
    },
    {
      "epoch": 0.1536170740786698,
      "grad_norm": 722.518310546875,
      "learning_rate": 9.923854974119434e-06,
      "loss": 125.0473,
      "step": 18570
    },
    {
      "epoch": 0.15369979732803904,
      "grad_norm": 1820.7373046875,
      "learning_rate": 9.92360957673457e-06,
      "loss": 141.7229,
      "step": 18580
    },
    {
      "epoch": 0.1537825205774083,
      "grad_norm": 1114.6781005859375,
      "learning_rate": 9.923363787600688e-06,
      "loss": 141.2934,
      "step": 18590
    },
    {
      "epoch": 0.15386524382677752,
      "grad_norm": 1381.0604248046875,
      "learning_rate": 9.923117606737347e-06,
      "loss": 116.5776,
      "step": 18600
    },
    {
      "epoch": 0.15394796707614675,
      "grad_norm": 1086.9346923828125,
      "learning_rate": 9.92287103416413e-06,
      "loss": 191.1018,
      "step": 18610
    },
    {
      "epoch": 0.15403069032551597,
      "grad_norm": 1572.17529296875,
      "learning_rate": 9.922624069900658e-06,
      "loss": 155.7499,
      "step": 18620
    },
    {
      "epoch": 0.15411341357488523,
      "grad_norm": 1132.931884765625,
      "learning_rate": 9.922376713966581e-06,
      "loss": 152.9908,
      "step": 18630
    },
    {
      "epoch": 0.15419613682425445,
      "grad_norm": 585.323486328125,
      "learning_rate": 9.92212896638158e-06,
      "loss": 152.2068,
      "step": 18640
    },
    {
      "epoch": 0.15427886007362368,
      "grad_norm": 595.2325439453125,
      "learning_rate": 9.921880827165367e-06,
      "loss": 118.7037,
      "step": 18650
    },
    {
      "epoch": 0.15436158332299293,
      "grad_norm": 4941.7626953125,
      "learning_rate": 9.921632296337683e-06,
      "loss": 153.8302,
      "step": 18660
    },
    {
      "epoch": 0.15444430657236216,
      "grad_norm": 1101.675048828125,
      "learning_rate": 9.921383373918305e-06,
      "loss": 180.7743,
      "step": 18670
    },
    {
      "epoch": 0.1545270298217314,
      "grad_norm": 1755.6380615234375,
      "learning_rate": 9.92113405992704e-06,
      "loss": 180.545,
      "step": 18680
    },
    {
      "epoch": 0.15460975307110064,
      "grad_norm": 925.7059326171875,
      "learning_rate": 9.92088435438372e-06,
      "loss": 134.731,
      "step": 18690
    },
    {
      "epoch": 0.15469247632046987,
      "grad_norm": 1003.4811401367188,
      "learning_rate": 9.920634257308217e-06,
      "loss": 123.1074,
      "step": 18700
    },
    {
      "epoch": 0.1547751995698391,
      "grad_norm": 853.6227416992188,
      "learning_rate": 9.920383768720429e-06,
      "loss": 150.079,
      "step": 18710
    },
    {
      "epoch": 0.15485792281920835,
      "grad_norm": 1113.7686767578125,
      "learning_rate": 9.920132888640286e-06,
      "loss": 155.0464,
      "step": 18720
    },
    {
      "epoch": 0.15494064606857758,
      "grad_norm": 1343.3956298828125,
      "learning_rate": 9.91988161708775e-06,
      "loss": 158.199,
      "step": 18730
    },
    {
      "epoch": 0.1550233693179468,
      "grad_norm": 704.8764038085938,
      "learning_rate": 9.919629954082813e-06,
      "loss": 153.2144,
      "step": 18740
    },
    {
      "epoch": 0.15510609256731606,
      "grad_norm": 550.6301879882812,
      "learning_rate": 9.919377899645497e-06,
      "loss": 141.5231,
      "step": 18750
    },
    {
      "epoch": 0.15518881581668528,
      "grad_norm": 658.1661376953125,
      "learning_rate": 9.91912545379586e-06,
      "loss": 138.1113,
      "step": 18760
    },
    {
      "epoch": 0.1552715390660545,
      "grad_norm": 1573.263916015625,
      "learning_rate": 9.918872616553986e-06,
      "loss": 129.6509,
      "step": 18770
    },
    {
      "epoch": 0.15535426231542376,
      "grad_norm": 1778.827392578125,
      "learning_rate": 9.918619387939991e-06,
      "loss": 155.8357,
      "step": 18780
    },
    {
      "epoch": 0.155436985564793,
      "grad_norm": 3056.092041015625,
      "learning_rate": 9.918365767974025e-06,
      "loss": 187.3279,
      "step": 18790
    },
    {
      "epoch": 0.15551970881416222,
      "grad_norm": 905.0897827148438,
      "learning_rate": 9.91811175667627e-06,
      "loss": 198.9183,
      "step": 18800
    },
    {
      "epoch": 0.15560243206353144,
      "grad_norm": 854.0653076171875,
      "learning_rate": 9.91785735406693e-06,
      "loss": 111.8965,
      "step": 18810
    },
    {
      "epoch": 0.1556851553129007,
      "grad_norm": 1693.5703125,
      "learning_rate": 9.917602560166253e-06,
      "loss": 138.9856,
      "step": 18820
    },
    {
      "epoch": 0.15576787856226992,
      "grad_norm": 1688.492431640625,
      "learning_rate": 9.917347374994507e-06,
      "loss": 118.313,
      "step": 18830
    },
    {
      "epoch": 0.15585060181163915,
      "grad_norm": 1094.8780517578125,
      "learning_rate": 9.917091798571998e-06,
      "loss": 122.0171,
      "step": 18840
    },
    {
      "epoch": 0.1559333250610084,
      "grad_norm": 1383.755859375,
      "learning_rate": 9.916835830919062e-06,
      "loss": 149.7231,
      "step": 18850
    },
    {
      "epoch": 0.15601604831037763,
      "grad_norm": 4172.68603515625,
      "learning_rate": 9.916579472056064e-06,
      "loss": 164.1563,
      "step": 18860
    },
    {
      "epoch": 0.15609877155974686,
      "grad_norm": 1112.282958984375,
      "learning_rate": 9.916322722003402e-06,
      "loss": 140.6031,
      "step": 18870
    },
    {
      "epoch": 0.1561814948091161,
      "grad_norm": 1334.9544677734375,
      "learning_rate": 9.916065580781504e-06,
      "loss": 125.8786,
      "step": 18880
    },
    {
      "epoch": 0.15626421805848534,
      "grad_norm": 759.095703125,
      "learning_rate": 9.91580804841083e-06,
      "loss": 123.02,
      "step": 18890
    },
    {
      "epoch": 0.15634694130785456,
      "grad_norm": 923.3683471679688,
      "learning_rate": 9.915550124911866e-06,
      "loss": 111.5178,
      "step": 18900
    },
    {
      "epoch": 0.15642966455722382,
      "grad_norm": 1883.232177734375,
      "learning_rate": 9.915291810305141e-06,
      "loss": 153.0945,
      "step": 18910
    },
    {
      "epoch": 0.15651238780659305,
      "grad_norm": 1571.4327392578125,
      "learning_rate": 9.915033104611204e-06,
      "loss": 152.6783,
      "step": 18920
    },
    {
      "epoch": 0.15659511105596227,
      "grad_norm": 842.9326171875,
      "learning_rate": 9.914774007850641e-06,
      "loss": 154.6972,
      "step": 18930
    },
    {
      "epoch": 0.15667783430533153,
      "grad_norm": 1248.7547607421875,
      "learning_rate": 9.914514520044065e-06,
      "loss": 169.4783,
      "step": 18940
    },
    {
      "epoch": 0.15676055755470075,
      "grad_norm": 976.1325073242188,
      "learning_rate": 9.914254641212124e-06,
      "loss": 114.911,
      "step": 18950
    },
    {
      "epoch": 0.15684328080406998,
      "grad_norm": 893.714111328125,
      "learning_rate": 9.913994371375494e-06,
      "loss": 81.8798,
      "step": 18960
    },
    {
      "epoch": 0.1569260040534392,
      "grad_norm": 1223.4085693359375,
      "learning_rate": 9.913733710554886e-06,
      "loss": 138.9431,
      "step": 18970
    },
    {
      "epoch": 0.15700872730280846,
      "grad_norm": 2076.80712890625,
      "learning_rate": 9.913472658771034e-06,
      "loss": 113.3516,
      "step": 18980
    },
    {
      "epoch": 0.15709145055217769,
      "grad_norm": 1146.1357421875,
      "learning_rate": 9.913211216044715e-06,
      "loss": 162.9254,
      "step": 18990
    },
    {
      "epoch": 0.1571741738015469,
      "grad_norm": 1230.30224609375,
      "learning_rate": 9.912949382396728e-06,
      "loss": 197.952,
      "step": 19000
    },
    {
      "epoch": 0.15725689705091617,
      "grad_norm": 1353.6494140625,
      "learning_rate": 9.912687157847905e-06,
      "loss": 137.2512,
      "step": 19010
    },
    {
      "epoch": 0.1573396203002854,
      "grad_norm": 1508.077392578125,
      "learning_rate": 9.91242454241911e-06,
      "loss": 133.2853,
      "step": 19020
    },
    {
      "epoch": 0.15742234354965462,
      "grad_norm": 935.4700927734375,
      "learning_rate": 9.912161536131242e-06,
      "loss": 126.4163,
      "step": 19030
    },
    {
      "epoch": 0.15750506679902387,
      "grad_norm": 920.0881958007812,
      "learning_rate": 9.911898139005222e-06,
      "loss": 106.0859,
      "step": 19040
    },
    {
      "epoch": 0.1575877900483931,
      "grad_norm": 1098.602783203125,
      "learning_rate": 9.91163435106201e-06,
      "loss": 113.0203,
      "step": 19050
    },
    {
      "epoch": 0.15767051329776233,
      "grad_norm": 529.7808227539062,
      "learning_rate": 9.911370172322595e-06,
      "loss": 100.4977,
      "step": 19060
    },
    {
      "epoch": 0.15775323654713158,
      "grad_norm": 1122.334228515625,
      "learning_rate": 9.911105602807996e-06,
      "loss": 147.1685,
      "step": 19070
    },
    {
      "epoch": 0.1578359597965008,
      "grad_norm": 1302.458740234375,
      "learning_rate": 9.910840642539261e-06,
      "loss": 138.237,
      "step": 19080
    },
    {
      "epoch": 0.15791868304587003,
      "grad_norm": 1324.593505859375,
      "learning_rate": 9.910575291537476e-06,
      "loss": 182.2281,
      "step": 19090
    },
    {
      "epoch": 0.1580014062952393,
      "grad_norm": 746.2387084960938,
      "learning_rate": 9.91030954982375e-06,
      "loss": 111.4794,
      "step": 19100
    },
    {
      "epoch": 0.15808412954460851,
      "grad_norm": 1335.113525390625,
      "learning_rate": 9.910043417419228e-06,
      "loss": 148.9087,
      "step": 19110
    },
    {
      "epoch": 0.15816685279397774,
      "grad_norm": 688.1320190429688,
      "learning_rate": 9.909776894345086e-06,
      "loss": 141.3004,
      "step": 19120
    },
    {
      "epoch": 0.158249576043347,
      "grad_norm": 1126.445068359375,
      "learning_rate": 9.909509980622532e-06,
      "loss": 112.016,
      "step": 19130
    },
    {
      "epoch": 0.15833229929271622,
      "grad_norm": 600.5185546875,
      "learning_rate": 9.909242676272797e-06,
      "loss": 114.159,
      "step": 19140
    },
    {
      "epoch": 0.15841502254208545,
      "grad_norm": 1174.3468017578125,
      "learning_rate": 9.908974981317155e-06,
      "loss": 171.2533,
      "step": 19150
    },
    {
      "epoch": 0.15849774579145468,
      "grad_norm": 795.6885986328125,
      "learning_rate": 9.9087068957769e-06,
      "loss": 136.8247,
      "step": 19160
    },
    {
      "epoch": 0.15858046904082393,
      "grad_norm": 1241.0509033203125,
      "learning_rate": 9.908438419673367e-06,
      "loss": 137.2768,
      "step": 19170
    },
    {
      "epoch": 0.15866319229019316,
      "grad_norm": 761.6776123046875,
      "learning_rate": 9.908169553027916e-06,
      "loss": 165.4491,
      "step": 19180
    },
    {
      "epoch": 0.15874591553956238,
      "grad_norm": 1572.6136474609375,
      "learning_rate": 9.90790029586194e-06,
      "loss": 124.7587,
      "step": 19190
    },
    {
      "epoch": 0.15882863878893164,
      "grad_norm": 732.7517700195312,
      "learning_rate": 9.907630648196857e-06,
      "loss": 142.462,
      "step": 19200
    },
    {
      "epoch": 0.15891136203830086,
      "grad_norm": 957.8698120117188,
      "learning_rate": 9.907360610054132e-06,
      "loss": 145.4445,
      "step": 19210
    },
    {
      "epoch": 0.1589940852876701,
      "grad_norm": 1933.4423828125,
      "learning_rate": 9.907090181455241e-06,
      "loss": 126.4228,
      "step": 19220
    },
    {
      "epoch": 0.15907680853703934,
      "grad_norm": 1341.31591796875,
      "learning_rate": 9.906819362421707e-06,
      "loss": 127.2506,
      "step": 19230
    },
    {
      "epoch": 0.15915953178640857,
      "grad_norm": 1401.2039794921875,
      "learning_rate": 9.906548152975076e-06,
      "loss": 142.5762,
      "step": 19240
    },
    {
      "epoch": 0.1592422550357778,
      "grad_norm": 1090.92578125,
      "learning_rate": 9.906276553136924e-06,
      "loss": 133.9682,
      "step": 19250
    },
    {
      "epoch": 0.15932497828514705,
      "grad_norm": 627.4381713867188,
      "learning_rate": 9.906004562928865e-06,
      "loss": 123.456,
      "step": 19260
    },
    {
      "epoch": 0.15940770153451628,
      "grad_norm": 1303.6290283203125,
      "learning_rate": 9.905732182372538e-06,
      "loss": 176.1459,
      "step": 19270
    },
    {
      "epoch": 0.1594904247838855,
      "grad_norm": 1230.2550048828125,
      "learning_rate": 9.905459411489617e-06,
      "loss": 150.8253,
      "step": 19280
    },
    {
      "epoch": 0.15957314803325476,
      "grad_norm": 557.597900390625,
      "learning_rate": 9.905186250301802e-06,
      "loss": 128.1924,
      "step": 19290
    },
    {
      "epoch": 0.15965587128262398,
      "grad_norm": 1013.2421875,
      "learning_rate": 9.904912698830828e-06,
      "loss": 148.6797,
      "step": 19300
    },
    {
      "epoch": 0.1597385945319932,
      "grad_norm": 1238.0384521484375,
      "learning_rate": 9.904638757098464e-06,
      "loss": 143.5567,
      "step": 19310
    },
    {
      "epoch": 0.15982131778136247,
      "grad_norm": 1117.36962890625,
      "learning_rate": 9.9043644251265e-06,
      "loss": 129.1499,
      "step": 19320
    },
    {
      "epoch": 0.1599040410307317,
      "grad_norm": 1082.633544921875,
      "learning_rate": 9.90408970293677e-06,
      "loss": 99.3771,
      "step": 19330
    },
    {
      "epoch": 0.15998676428010092,
      "grad_norm": 1239.34326171875,
      "learning_rate": 9.903814590551127e-06,
      "loss": 152.1191,
      "step": 19340
    },
    {
      "epoch": 0.16006948752947014,
      "grad_norm": 1026.2008056640625,
      "learning_rate": 9.903539087991462e-06,
      "loss": 138.5603,
      "step": 19350
    },
    {
      "epoch": 0.1601522107788394,
      "grad_norm": 716.8160400390625,
      "learning_rate": 9.903263195279698e-06,
      "loss": 121.6254,
      "step": 19360
    },
    {
      "epoch": 0.16023493402820863,
      "grad_norm": 1355.804931640625,
      "learning_rate": 9.902986912437784e-06,
      "loss": 121.697,
      "step": 19370
    },
    {
      "epoch": 0.16031765727757785,
      "grad_norm": 1001.9555053710938,
      "learning_rate": 9.902710239487702e-06,
      "loss": 123.8956,
      "step": 19380
    },
    {
      "epoch": 0.1604003805269471,
      "grad_norm": 929.7130737304688,
      "learning_rate": 9.902433176451466e-06,
      "loss": 108.8211,
      "step": 19390
    },
    {
      "epoch": 0.16048310377631633,
      "grad_norm": 1794.6314697265625,
      "learning_rate": 9.902155723351124e-06,
      "loss": 119.6667,
      "step": 19400
    },
    {
      "epoch": 0.16056582702568556,
      "grad_norm": 981.8839721679688,
      "learning_rate": 9.901877880208747e-06,
      "loss": 123.9001,
      "step": 19410
    },
    {
      "epoch": 0.1606485502750548,
      "grad_norm": 1454.4476318359375,
      "learning_rate": 9.901599647046443e-06,
      "loss": 131.0193,
      "step": 19420
    },
    {
      "epoch": 0.16073127352442404,
      "grad_norm": 1284.30224609375,
      "learning_rate": 9.901321023886351e-06,
      "loss": 169.1719,
      "step": 19430
    },
    {
      "epoch": 0.16081399677379327,
      "grad_norm": 1159.77783203125,
      "learning_rate": 9.901042010750641e-06,
      "loss": 100.9739,
      "step": 19440
    },
    {
      "epoch": 0.16089672002316252,
      "grad_norm": 1844.110107421875,
      "learning_rate": 9.900762607661509e-06,
      "loss": 153.9659,
      "step": 19450
    },
    {
      "epoch": 0.16097944327253175,
      "grad_norm": 701.9683227539062,
      "learning_rate": 9.900482814641188e-06,
      "loss": 109.9286,
      "step": 19460
    },
    {
      "epoch": 0.16106216652190097,
      "grad_norm": 1962.1533203125,
      "learning_rate": 9.90020263171194e-06,
      "loss": 158.6689,
      "step": 19470
    },
    {
      "epoch": 0.16114488977127023,
      "grad_norm": 1527.3931884765625,
      "learning_rate": 9.899922058896058e-06,
      "loss": 129.6219,
      "step": 19480
    },
    {
      "epoch": 0.16122761302063945,
      "grad_norm": 935.3746337890625,
      "learning_rate": 9.899641096215865e-06,
      "loss": 187.1026,
      "step": 19490
    },
    {
      "epoch": 0.16131033627000868,
      "grad_norm": 1236.936279296875,
      "learning_rate": 9.899359743693715e-06,
      "loss": 194.122,
      "step": 19500
    },
    {
      "epoch": 0.16139305951937793,
      "grad_norm": 2321.50439453125,
      "learning_rate": 9.899078001351996e-06,
      "loss": 164.5937,
      "step": 19510
    },
    {
      "epoch": 0.16147578276874716,
      "grad_norm": 1305.2666015625,
      "learning_rate": 9.898795869213125e-06,
      "loss": 149.4349,
      "step": 19520
    },
    {
      "epoch": 0.1615585060181164,
      "grad_norm": 1175.187744140625,
      "learning_rate": 9.898513347299549e-06,
      "loss": 142.7042,
      "step": 19530
    },
    {
      "epoch": 0.16164122926748561,
      "grad_norm": 1031.960693359375,
      "learning_rate": 9.898230435633747e-06,
      "loss": 151.7943,
      "step": 19540
    },
    {
      "epoch": 0.16172395251685487,
      "grad_norm": 1225.9884033203125,
      "learning_rate": 9.897947134238228e-06,
      "loss": 162.5945,
      "step": 19550
    },
    {
      "epoch": 0.1618066757662241,
      "grad_norm": 1006.8120727539062,
      "learning_rate": 9.897663443135534e-06,
      "loss": 127.9761,
      "step": 19560
    },
    {
      "epoch": 0.16188939901559332,
      "grad_norm": 2127.416748046875,
      "learning_rate": 9.897379362348239e-06,
      "loss": 171.9894,
      "step": 19570
    },
    {
      "epoch": 0.16197212226496258,
      "grad_norm": 745.5164794921875,
      "learning_rate": 9.897094891898942e-06,
      "loss": 150.5477,
      "step": 19580
    },
    {
      "epoch": 0.1620548455143318,
      "grad_norm": 2045.43896484375,
      "learning_rate": 9.89681003181028e-06,
      "loss": 143.4611,
      "step": 19590
    },
    {
      "epoch": 0.16213756876370103,
      "grad_norm": 1801.7623291015625,
      "learning_rate": 9.896524782104917e-06,
      "loss": 136.5524,
      "step": 19600
    },
    {
      "epoch": 0.16222029201307028,
      "grad_norm": 865.27001953125,
      "learning_rate": 9.89623914280555e-06,
      "loss": 135.0138,
      "step": 19610
    },
    {
      "epoch": 0.1623030152624395,
      "grad_norm": 1108.943359375,
      "learning_rate": 9.895953113934904e-06,
      "loss": 131.2855,
      "step": 19620
    },
    {
      "epoch": 0.16238573851180874,
      "grad_norm": 1284.2874755859375,
      "learning_rate": 9.895666695515739e-06,
      "loss": 158.5307,
      "step": 19630
    },
    {
      "epoch": 0.162468461761178,
      "grad_norm": 1160.2169189453125,
      "learning_rate": 9.895379887570842e-06,
      "loss": 146.816,
      "step": 19640
    },
    {
      "epoch": 0.16255118501054722,
      "grad_norm": 1387.5469970703125,
      "learning_rate": 9.895092690123036e-06,
      "loss": 130.8016,
      "step": 19650
    },
    {
      "epoch": 0.16263390825991644,
      "grad_norm": 1327.9295654296875,
      "learning_rate": 9.894805103195168e-06,
      "loss": 131.4063,
      "step": 19660
    },
    {
      "epoch": 0.1627166315092857,
      "grad_norm": 1514.6529541015625,
      "learning_rate": 9.894517126810122e-06,
      "loss": 209.5621,
      "step": 19670
    },
    {
      "epoch": 0.16279935475865492,
      "grad_norm": 2022.22021484375,
      "learning_rate": 9.894228760990811e-06,
      "loss": 152.1554,
      "step": 19680
    },
    {
      "epoch": 0.16288207800802415,
      "grad_norm": 1740.86767578125,
      "learning_rate": 9.893940005760181e-06,
      "loss": 154.0035,
      "step": 19690
    },
    {
      "epoch": 0.16296480125739338,
      "grad_norm": 1580.342529296875,
      "learning_rate": 9.893650861141204e-06,
      "loss": 157.6928,
      "step": 19700
    },
    {
      "epoch": 0.16304752450676263,
      "grad_norm": 1006.531494140625,
      "learning_rate": 9.893361327156887e-06,
      "loss": 127.0846,
      "step": 19710
    },
    {
      "epoch": 0.16313024775613186,
      "grad_norm": 1319.7847900390625,
      "learning_rate": 9.893071403830265e-06,
      "loss": 136.8425,
      "step": 19720
    },
    {
      "epoch": 0.16321297100550108,
      "grad_norm": 830.5723876953125,
      "learning_rate": 9.892781091184409e-06,
      "loss": 136.2878,
      "step": 19730
    },
    {
      "epoch": 0.16329569425487034,
      "grad_norm": 759.5004272460938,
      "learning_rate": 9.892490389242417e-06,
      "loss": 120.3061,
      "step": 19740
    },
    {
      "epoch": 0.16337841750423956,
      "grad_norm": 2213.310546875,
      "learning_rate": 9.892199298027416e-06,
      "loss": 143.1016,
      "step": 19750
    },
    {
      "epoch": 0.1634611407536088,
      "grad_norm": 1020.21337890625,
      "learning_rate": 9.891907817562572e-06,
      "loss": 116.1548,
      "step": 19760
    },
    {
      "epoch": 0.16354386400297805,
      "grad_norm": 810.1748657226562,
      "learning_rate": 9.891615947871072e-06,
      "loss": 141.0581,
      "step": 19770
    },
    {
      "epoch": 0.16362658725234727,
      "grad_norm": 810.9425048828125,
      "learning_rate": 9.89132368897614e-06,
      "loss": 130.3051,
      "step": 19780
    },
    {
      "epoch": 0.1637093105017165,
      "grad_norm": 1321.05908203125,
      "learning_rate": 9.891031040901031e-06,
      "loss": 154.2215,
      "step": 19790
    },
    {
      "epoch": 0.16379203375108575,
      "grad_norm": 1216.8099365234375,
      "learning_rate": 9.890738003669029e-06,
      "loss": 164.4314,
      "step": 19800
    },
    {
      "epoch": 0.16387475700045498,
      "grad_norm": 2961.08447265625,
      "learning_rate": 9.890444577303448e-06,
      "loss": 184.4128,
      "step": 19810
    },
    {
      "epoch": 0.1639574802498242,
      "grad_norm": 702.2813110351562,
      "learning_rate": 9.890150761827639e-06,
      "loss": 118.5094,
      "step": 19820
    },
    {
      "epoch": 0.16404020349919346,
      "grad_norm": 1224.574951171875,
      "learning_rate": 9.889856557264975e-06,
      "loss": 164.4189,
      "step": 19830
    },
    {
      "epoch": 0.1641229267485627,
      "grad_norm": 1510.7064208984375,
      "learning_rate": 9.889561963638866e-06,
      "loss": 168.8556,
      "step": 19840
    },
    {
      "epoch": 0.1642056499979319,
      "grad_norm": 1308.8349609375,
      "learning_rate": 9.889266980972752e-06,
      "loss": 157.53,
      "step": 19850
    },
    {
      "epoch": 0.16428837324730117,
      "grad_norm": 554.7841796875,
      "learning_rate": 9.888971609290103e-06,
      "loss": 123.5679,
      "step": 19860
    },
    {
      "epoch": 0.1643710964966704,
      "grad_norm": 1066.4405517578125,
      "learning_rate": 9.88867584861442e-06,
      "loss": 115.5183,
      "step": 19870
    },
    {
      "epoch": 0.16445381974603962,
      "grad_norm": 823.9727172851562,
      "learning_rate": 9.888379698969236e-06,
      "loss": 127.0505,
      "step": 19880
    },
    {
      "epoch": 0.16453654299540885,
      "grad_norm": 2073.93017578125,
      "learning_rate": 9.888083160378114e-06,
      "loss": 142.6533,
      "step": 19890
    },
    {
      "epoch": 0.1646192662447781,
      "grad_norm": 2434.051513671875,
      "learning_rate": 9.887786232864648e-06,
      "loss": 147.1622,
      "step": 19900
    },
    {
      "epoch": 0.16470198949414733,
      "grad_norm": 724.6605224609375,
      "learning_rate": 9.887488916452463e-06,
      "loss": 121.8898,
      "step": 19910
    },
    {
      "epoch": 0.16478471274351655,
      "grad_norm": 1281.83203125,
      "learning_rate": 9.887191211165217e-06,
      "loss": 151.6535,
      "step": 19920
    },
    {
      "epoch": 0.1648674359928858,
      "grad_norm": 1131.1641845703125,
      "learning_rate": 9.886893117026593e-06,
      "loss": 149.5577,
      "step": 19930
    },
    {
      "epoch": 0.16495015924225503,
      "grad_norm": 1077.4385986328125,
      "learning_rate": 9.886594634060314e-06,
      "loss": 203.1148,
      "step": 19940
    },
    {
      "epoch": 0.16503288249162426,
      "grad_norm": 1563.59228515625,
      "learning_rate": 9.886295762290125e-06,
      "loss": 156.8315,
      "step": 19950
    },
    {
      "epoch": 0.16511560574099352,
      "grad_norm": 728.146240234375,
      "learning_rate": 9.885996501739808e-06,
      "loss": 123.6347,
      "step": 19960
    },
    {
      "epoch": 0.16519832899036274,
      "grad_norm": 1174.09521484375,
      "learning_rate": 9.885696852433174e-06,
      "loss": 171.4022,
      "step": 19970
    },
    {
      "epoch": 0.16528105223973197,
      "grad_norm": 2437.55908203125,
      "learning_rate": 9.885396814394062e-06,
      "loss": 166.2973,
      "step": 19980
    },
    {
      "epoch": 0.16536377548910122,
      "grad_norm": 756.368896484375,
      "learning_rate": 9.885096387646346e-06,
      "loss": 102.5183,
      "step": 19990
    },
    {
      "epoch": 0.16544649873847045,
      "grad_norm": 1208.659423828125,
      "learning_rate": 9.88479557221393e-06,
      "loss": 146.3919,
      "step": 20000
    },
    {
      "epoch": 0.16552922198783968,
      "grad_norm": 793.578857421875,
      "learning_rate": 9.88449436812075e-06,
      "loss": 151.3374,
      "step": 20010
    },
    {
      "epoch": 0.16561194523720893,
      "grad_norm": 3023.392333984375,
      "learning_rate": 9.88419277539077e-06,
      "loss": 147.2389,
      "step": 20020
    },
    {
      "epoch": 0.16569466848657816,
      "grad_norm": 1020.14404296875,
      "learning_rate": 9.883890794047985e-06,
      "loss": 133.7473,
      "step": 20030
    },
    {
      "epoch": 0.16577739173594738,
      "grad_norm": 925.8684692382812,
      "learning_rate": 9.883588424116424e-06,
      "loss": 145.2095,
      "step": 20040
    },
    {
      "epoch": 0.16586011498531664,
      "grad_norm": 1657.950927734375,
      "learning_rate": 9.883285665620145e-06,
      "loss": 131.4692,
      "step": 20050
    },
    {
      "epoch": 0.16594283823468586,
      "grad_norm": 1638.5106201171875,
      "learning_rate": 9.882982518583238e-06,
      "loss": 120.6384,
      "step": 20060
    },
    {
      "epoch": 0.1660255614840551,
      "grad_norm": 994.8275146484375,
      "learning_rate": 9.882678983029819e-06,
      "loss": 191.7884,
      "step": 20070
    },
    {
      "epoch": 0.16610828473342432,
      "grad_norm": 1858.4609375,
      "learning_rate": 9.882375058984044e-06,
      "loss": 145.8128,
      "step": 20080
    },
    {
      "epoch": 0.16619100798279357,
      "grad_norm": 640.3125,
      "learning_rate": 9.882070746470092e-06,
      "loss": 113.2083,
      "step": 20090
    },
    {
      "epoch": 0.1662737312321628,
      "grad_norm": 1469.511474609375,
      "learning_rate": 9.881766045512176e-06,
      "loss": 189.0106,
      "step": 20100
    },
    {
      "epoch": 0.16635645448153202,
      "grad_norm": 740.4965209960938,
      "learning_rate": 9.88146095613454e-06,
      "loss": 130.8047,
      "step": 20110
    },
    {
      "epoch": 0.16643917773090128,
      "grad_norm": 683.9896240234375,
      "learning_rate": 9.881155478361459e-06,
      "loss": 175.5372,
      "step": 20120
    },
    {
      "epoch": 0.1665219009802705,
      "grad_norm": 938.4227905273438,
      "learning_rate": 9.880849612217238e-06,
      "loss": 108.5235,
      "step": 20130
    },
    {
      "epoch": 0.16660462422963973,
      "grad_norm": 841.4732666015625,
      "learning_rate": 9.880543357726214e-06,
      "loss": 142.9208,
      "step": 20140
    },
    {
      "epoch": 0.16668734747900898,
      "grad_norm": 764.7952880859375,
      "learning_rate": 9.880236714912754e-06,
      "loss": 136.3933,
      "step": 20150
    },
    {
      "epoch": 0.1667700707283782,
      "grad_norm": 1391.5673828125,
      "learning_rate": 9.879929683801254e-06,
      "loss": 138.4007,
      "step": 20160
    },
    {
      "epoch": 0.16685279397774744,
      "grad_norm": 1550.94873046875,
      "learning_rate": 9.879622264416147e-06,
      "loss": 147.9795,
      "step": 20170
    },
    {
      "epoch": 0.1669355172271167,
      "grad_norm": 879.2140502929688,
      "learning_rate": 9.87931445678189e-06,
      "loss": 155.5872,
      "step": 20180
    },
    {
      "epoch": 0.16701824047648592,
      "grad_norm": 584.1538696289062,
      "learning_rate": 9.879006260922975e-06,
      "loss": 98.7441,
      "step": 20190
    },
    {
      "epoch": 0.16710096372585515,
      "grad_norm": 719.3741455078125,
      "learning_rate": 9.878697676863922e-06,
      "loss": 126.3837,
      "step": 20200
    },
    {
      "epoch": 0.1671836869752244,
      "grad_norm": 1768.941162109375,
      "learning_rate": 9.878388704629286e-06,
      "loss": 181.439,
      "step": 20210
    },
    {
      "epoch": 0.16726641022459363,
      "grad_norm": 1062.1995849609375,
      "learning_rate": 9.87807934424365e-06,
      "loss": 155.1683,
      "step": 20220
    },
    {
      "epoch": 0.16734913347396285,
      "grad_norm": 1074.44482421875,
      "learning_rate": 9.877769595731629e-06,
      "loss": 149.4426,
      "step": 20230
    },
    {
      "epoch": 0.16743185672333208,
      "grad_norm": 766.1312866210938,
      "learning_rate": 9.877459459117864e-06,
      "loss": 133.7859,
      "step": 20240
    },
    {
      "epoch": 0.16751457997270133,
      "grad_norm": 1180.6207275390625,
      "learning_rate": 9.877148934427037e-06,
      "loss": 185.559,
      "step": 20250
    },
    {
      "epoch": 0.16759730322207056,
      "grad_norm": 936.2619018554688,
      "learning_rate": 9.87683802168385e-06,
      "loss": 153.8027,
      "step": 20260
    },
    {
      "epoch": 0.16768002647143979,
      "grad_norm": 1289.442626953125,
      "learning_rate": 9.876526720913045e-06,
      "loss": 146.8949,
      "step": 20270
    },
    {
      "epoch": 0.16776274972080904,
      "grad_norm": 1198.4373779296875,
      "learning_rate": 9.87621503213939e-06,
      "loss": 101.3234,
      "step": 20280
    },
    {
      "epoch": 0.16784547297017827,
      "grad_norm": 1139.7901611328125,
      "learning_rate": 9.875902955387682e-06,
      "loss": 105.7266,
      "step": 20290
    },
    {
      "epoch": 0.1679281962195475,
      "grad_norm": 885.1135864257812,
      "learning_rate": 9.875590490682754e-06,
      "loss": 139.6578,
      "step": 20300
    },
    {
      "epoch": 0.16801091946891675,
      "grad_norm": 1269.400146484375,
      "learning_rate": 9.875277638049466e-06,
      "loss": 148.561,
      "step": 20310
    },
    {
      "epoch": 0.16809364271828597,
      "grad_norm": 1671.2281494140625,
      "learning_rate": 9.87496439751271e-06,
      "loss": 157.8267,
      "step": 20320
    },
    {
      "epoch": 0.1681763659676552,
      "grad_norm": 913.9154663085938,
      "learning_rate": 9.87465076909741e-06,
      "loss": 141.2827,
      "step": 20330
    },
    {
      "epoch": 0.16825908921702445,
      "grad_norm": 1082.798583984375,
      "learning_rate": 9.874336752828523e-06,
      "loss": 150.5321,
      "step": 20340
    },
    {
      "epoch": 0.16834181246639368,
      "grad_norm": 967.8886108398438,
      "learning_rate": 9.87402234873103e-06,
      "loss": 142.4086,
      "step": 20350
    },
    {
      "epoch": 0.1684245357157629,
      "grad_norm": 1056.4305419921875,
      "learning_rate": 9.873707556829945e-06,
      "loss": 97.3924,
      "step": 20360
    },
    {
      "epoch": 0.16850725896513216,
      "grad_norm": 837.5071411132812,
      "learning_rate": 9.873392377150318e-06,
      "loss": 123.7359,
      "step": 20370
    },
    {
      "epoch": 0.1685899822145014,
      "grad_norm": 874.9818115234375,
      "learning_rate": 9.873076809717226e-06,
      "loss": 109.1823,
      "step": 20380
    },
    {
      "epoch": 0.16867270546387061,
      "grad_norm": 727.0296020507812,
      "learning_rate": 9.872760854555776e-06,
      "loss": 143.6749,
      "step": 20390
    },
    {
      "epoch": 0.16875542871323987,
      "grad_norm": 1192.26904296875,
      "learning_rate": 9.872444511691108e-06,
      "loss": 128.2298,
      "step": 20400
    },
    {
      "epoch": 0.1688381519626091,
      "grad_norm": 765.3291015625,
      "learning_rate": 9.872127781148392e-06,
      "loss": 140.1519,
      "step": 20410
    },
    {
      "epoch": 0.16892087521197832,
      "grad_norm": 1048.41064453125,
      "learning_rate": 9.871810662952828e-06,
      "loss": 153.647,
      "step": 20420
    },
    {
      "epoch": 0.16900359846134755,
      "grad_norm": 1050.804931640625,
      "learning_rate": 9.87149315712965e-06,
      "loss": 160.4528,
      "step": 20430
    },
    {
      "epoch": 0.1690863217107168,
      "grad_norm": 1345.5584716796875,
      "learning_rate": 9.871175263704116e-06,
      "loss": 119.833,
      "step": 20440
    },
    {
      "epoch": 0.16916904496008603,
      "grad_norm": 1076.057861328125,
      "learning_rate": 9.870856982701522e-06,
      "loss": 123.713,
      "step": 20450
    },
    {
      "epoch": 0.16925176820945526,
      "grad_norm": 1091.8779296875,
      "learning_rate": 9.870538314147194e-06,
      "loss": 157.7124,
      "step": 20460
    },
    {
      "epoch": 0.1693344914588245,
      "grad_norm": 1105.4384765625,
      "learning_rate": 9.870219258066485e-06,
      "loss": 165.7277,
      "step": 20470
    },
    {
      "epoch": 0.16941721470819374,
      "grad_norm": 1101.884033203125,
      "learning_rate": 9.86989981448478e-06,
      "loss": 93.6696,
      "step": 20480
    },
    {
      "epoch": 0.16949993795756296,
      "grad_norm": 1253.4547119140625,
      "learning_rate": 9.869579983427497e-06,
      "loss": 156.5219,
      "step": 20490
    },
    {
      "epoch": 0.16958266120693222,
      "grad_norm": 571.4030151367188,
      "learning_rate": 9.869259764920081e-06,
      "loss": 108.441,
      "step": 20500
    },
    {
      "epoch": 0.16966538445630144,
      "grad_norm": 1125.4053955078125,
      "learning_rate": 9.868939158988016e-06,
      "loss": 111.0379,
      "step": 20510
    },
    {
      "epoch": 0.16974810770567067,
      "grad_norm": 1643.4638671875,
      "learning_rate": 9.868618165656805e-06,
      "loss": 174.5399,
      "step": 20520
    },
    {
      "epoch": 0.16983083095503992,
      "grad_norm": 1219.7093505859375,
      "learning_rate": 9.868296784951992e-06,
      "loss": 143.0416,
      "step": 20530
    },
    {
      "epoch": 0.16991355420440915,
      "grad_norm": 716.170654296875,
      "learning_rate": 9.867975016899145e-06,
      "loss": 129.3315,
      "step": 20540
    },
    {
      "epoch": 0.16999627745377838,
      "grad_norm": 722.2046508789062,
      "learning_rate": 9.867652861523866e-06,
      "loss": 110.3743,
      "step": 20550
    },
    {
      "epoch": 0.17007900070314763,
      "grad_norm": 8400.3681640625,
      "learning_rate": 9.86733031885179e-06,
      "loss": 293.1446,
      "step": 20560
    },
    {
      "epoch": 0.17016172395251686,
      "grad_norm": 1231.3631591796875,
      "learning_rate": 9.867007388908579e-06,
      "loss": 158.7573,
      "step": 20570
    },
    {
      "epoch": 0.17024444720188608,
      "grad_norm": 813.4060668945312,
      "learning_rate": 9.866684071719926e-06,
      "loss": 114.0276,
      "step": 20580
    },
    {
      "epoch": 0.17032717045125534,
      "grad_norm": 691.1693115234375,
      "learning_rate": 9.866360367311557e-06,
      "loss": 124.2069,
      "step": 20590
    },
    {
      "epoch": 0.17040989370062457,
      "grad_norm": 1228.2000732421875,
      "learning_rate": 9.866036275709226e-06,
      "loss": 113.3982,
      "step": 20600
    },
    {
      "epoch": 0.1704926169499938,
      "grad_norm": 1183.33935546875,
      "learning_rate": 9.86571179693872e-06,
      "loss": 124.6813,
      "step": 20610
    },
    {
      "epoch": 0.17057534019936302,
      "grad_norm": 1132.4244384765625,
      "learning_rate": 9.865386931025858e-06,
      "loss": 109.3566,
      "step": 20620
    },
    {
      "epoch": 0.17065806344873227,
      "grad_norm": 536.39453125,
      "learning_rate": 9.865061677996487e-06,
      "loss": 94.1924,
      "step": 20630
    },
    {
      "epoch": 0.1707407866981015,
      "grad_norm": 1061.803955078125,
      "learning_rate": 9.864736037876487e-06,
      "loss": 96.9849,
      "step": 20640
    },
    {
      "epoch": 0.17082350994747073,
      "grad_norm": 1086.843505859375,
      "learning_rate": 9.864410010691766e-06,
      "loss": 158.8272,
      "step": 20650
    },
    {
      "epoch": 0.17090623319683998,
      "grad_norm": 1095.378662109375,
      "learning_rate": 9.864083596468263e-06,
      "loss": 165.8036,
      "step": 20660
    },
    {
      "epoch": 0.1709889564462092,
      "grad_norm": 773.133544921875,
      "learning_rate": 9.863756795231953e-06,
      "loss": 107.4877,
      "step": 20670
    },
    {
      "epoch": 0.17107167969557843,
      "grad_norm": 922.110107421875,
      "learning_rate": 9.863429607008837e-06,
      "loss": 152.2869,
      "step": 20680
    },
    {
      "epoch": 0.1711544029449477,
      "grad_norm": 828.44677734375,
      "learning_rate": 9.863102031824946e-06,
      "loss": 103.4225,
      "step": 20690
    },
    {
      "epoch": 0.1712371261943169,
      "grad_norm": 1600.447021484375,
      "learning_rate": 9.862774069706346e-06,
      "loss": 116.9802,
      "step": 20700
    },
    {
      "epoch": 0.17131984944368614,
      "grad_norm": 1819.4007568359375,
      "learning_rate": 9.86244572067913e-06,
      "loss": 193.74,
      "step": 20710
    },
    {
      "epoch": 0.1714025726930554,
      "grad_norm": 1109.385009765625,
      "learning_rate": 9.862116984769424e-06,
      "loss": 155.2172,
      "step": 20720
    },
    {
      "epoch": 0.17148529594242462,
      "grad_norm": 1088.8150634765625,
      "learning_rate": 9.861787862003384e-06,
      "loss": 102.3083,
      "step": 20730
    },
    {
      "epoch": 0.17156801919179385,
      "grad_norm": 863.4269409179688,
      "learning_rate": 9.861458352407196e-06,
      "loss": 163.3116,
      "step": 20740
    },
    {
      "epoch": 0.1716507424411631,
      "grad_norm": 985.338623046875,
      "learning_rate": 9.861128456007076e-06,
      "loss": 128.2076,
      "step": 20750
    },
    {
      "epoch": 0.17173346569053233,
      "grad_norm": 2616.19189453125,
      "learning_rate": 9.860798172829277e-06,
      "loss": 128.8503,
      "step": 20760
    },
    {
      "epoch": 0.17181618893990155,
      "grad_norm": 821.7667846679688,
      "learning_rate": 9.860467502900076e-06,
      "loss": 139.1303,
      "step": 20770
    },
    {
      "epoch": 0.1718989121892708,
      "grad_norm": 1093.295654296875,
      "learning_rate": 9.860136446245779e-06,
      "loss": 158.2073,
      "step": 20780
    },
    {
      "epoch": 0.17198163543864003,
      "grad_norm": 2161.525390625,
      "learning_rate": 9.859805002892733e-06,
      "loss": 159.4854,
      "step": 20790
    },
    {
      "epoch": 0.17206435868800926,
      "grad_norm": 936.6251831054688,
      "learning_rate": 9.859473172867304e-06,
      "loss": 150.7438,
      "step": 20800
    },
    {
      "epoch": 0.1721470819373785,
      "grad_norm": 735.8521118164062,
      "learning_rate": 9.859140956195898e-06,
      "loss": 166.0925,
      "step": 20810
    },
    {
      "epoch": 0.17222980518674774,
      "grad_norm": 1382.367431640625,
      "learning_rate": 9.858808352904946e-06,
      "loss": 163.85,
      "step": 20820
    },
    {
      "epoch": 0.17231252843611697,
      "grad_norm": 1131.4239501953125,
      "learning_rate": 9.858475363020913e-06,
      "loss": 138.0052,
      "step": 20830
    },
    {
      "epoch": 0.1723952516854862,
      "grad_norm": 1062.67431640625,
      "learning_rate": 9.858141986570294e-06,
      "loss": 127.1865,
      "step": 20840
    },
    {
      "epoch": 0.17247797493485545,
      "grad_norm": 790.3716430664062,
      "learning_rate": 9.85780822357961e-06,
      "loss": 99.19,
      "step": 20850
    },
    {
      "epoch": 0.17256069818422468,
      "grad_norm": 1042.801025390625,
      "learning_rate": 9.857474074075422e-06,
      "loss": 152.4688,
      "step": 20860
    },
    {
      "epoch": 0.1726434214335939,
      "grad_norm": 872.380126953125,
      "learning_rate": 9.857139538084313e-06,
      "loss": 180.6786,
      "step": 20870
    },
    {
      "epoch": 0.17272614468296316,
      "grad_norm": 873.7167358398438,
      "learning_rate": 9.856804615632904e-06,
      "loss": 153.8926,
      "step": 20880
    },
    {
      "epoch": 0.17280886793233238,
      "grad_norm": 1261.3304443359375,
      "learning_rate": 9.85646930674784e-06,
      "loss": 138.4761,
      "step": 20890
    },
    {
      "epoch": 0.1728915911817016,
      "grad_norm": 1224.0684814453125,
      "learning_rate": 9.856133611455802e-06,
      "loss": 116.0446,
      "step": 20900
    },
    {
      "epoch": 0.17297431443107086,
      "grad_norm": 1034.66552734375,
      "learning_rate": 9.855797529783499e-06,
      "loss": 143.5475,
      "step": 20910
    },
    {
      "epoch": 0.1730570376804401,
      "grad_norm": 1428.3978271484375,
      "learning_rate": 9.855461061757673e-06,
      "loss": 202.7229,
      "step": 20920
    },
    {
      "epoch": 0.17313976092980932,
      "grad_norm": 1692.571533203125,
      "learning_rate": 9.855124207405093e-06,
      "loss": 129.3957,
      "step": 20930
    },
    {
      "epoch": 0.17322248417917857,
      "grad_norm": 901.4539184570312,
      "learning_rate": 9.854786966752561e-06,
      "loss": 114.9998,
      "step": 20940
    },
    {
      "epoch": 0.1733052074285478,
      "grad_norm": 1734.4339599609375,
      "learning_rate": 9.854449339826912e-06,
      "loss": 126.6563,
      "step": 20950
    },
    {
      "epoch": 0.17338793067791702,
      "grad_norm": 936.6329956054688,
      "learning_rate": 9.854111326655006e-06,
      "loss": 148.8187,
      "step": 20960
    },
    {
      "epoch": 0.17347065392728625,
      "grad_norm": 792.8075561523438,
      "learning_rate": 9.85377292726374e-06,
      "loss": 140.2986,
      "step": 20970
    },
    {
      "epoch": 0.1735533771766555,
      "grad_norm": 852.9043579101562,
      "learning_rate": 9.85343414168004e-06,
      "loss": 142.9736,
      "step": 20980
    },
    {
      "epoch": 0.17363610042602473,
      "grad_norm": 615.2127685546875,
      "learning_rate": 9.853094969930857e-06,
      "loss": 131.7546,
      "step": 20990
    },
    {
      "epoch": 0.17371882367539396,
      "grad_norm": 1290.294921875,
      "learning_rate": 9.85275541204318e-06,
      "loss": 157.3864,
      "step": 21000
    },
    {
      "epoch": 0.1738015469247632,
      "grad_norm": 428.7129821777344,
      "learning_rate": 9.852415468044027e-06,
      "loss": 117.3043,
      "step": 21010
    },
    {
      "epoch": 0.17388427017413244,
      "grad_norm": 1278.5567626953125,
      "learning_rate": 9.852075137960446e-06,
      "loss": 136.686,
      "step": 21020
    },
    {
      "epoch": 0.17396699342350166,
      "grad_norm": 1619.6778564453125,
      "learning_rate": 9.851734421819511e-06,
      "loss": 136.3727,
      "step": 21030
    },
    {
      "epoch": 0.17404971667287092,
      "grad_norm": 1088.0845947265625,
      "learning_rate": 9.851393319648338e-06,
      "loss": 114.8988,
      "step": 21040
    },
    {
      "epoch": 0.17413243992224015,
      "grad_norm": 738.9354858398438,
      "learning_rate": 9.851051831474062e-06,
      "loss": 138.9047,
      "step": 21050
    },
    {
      "epoch": 0.17421516317160937,
      "grad_norm": 926.2405395507812,
      "learning_rate": 9.850709957323855e-06,
      "loss": 127.3761,
      "step": 21060
    },
    {
      "epoch": 0.17429788642097863,
      "grad_norm": 911.6777954101562,
      "learning_rate": 9.85036769722492e-06,
      "loss": 163.0244,
      "step": 21070
    },
    {
      "epoch": 0.17438060967034785,
      "grad_norm": 868.7709350585938,
      "learning_rate": 9.850025051204484e-06,
      "loss": 139.7337,
      "step": 21080
    },
    {
      "epoch": 0.17446333291971708,
      "grad_norm": 810.2794189453125,
      "learning_rate": 9.849682019289816e-06,
      "loss": 129.7191,
      "step": 21090
    },
    {
      "epoch": 0.17454605616908633,
      "grad_norm": 862.4880981445312,
      "learning_rate": 9.849338601508204e-06,
      "loss": 110.1159,
      "step": 21100
    },
    {
      "epoch": 0.17462877941845556,
      "grad_norm": 1751.4161376953125,
      "learning_rate": 9.848994797886978e-06,
      "loss": 162.2478,
      "step": 21110
    },
    {
      "epoch": 0.1747115026678248,
      "grad_norm": 961.6451416015625,
      "learning_rate": 9.84865060845349e-06,
      "loss": 151.0574,
      "step": 21120
    },
    {
      "epoch": 0.17479422591719404,
      "grad_norm": 984.0369873046875,
      "learning_rate": 9.848306033235123e-06,
      "loss": 114.3529,
      "step": 21130
    },
    {
      "epoch": 0.17487694916656327,
      "grad_norm": 1510.6654052734375,
      "learning_rate": 9.847961072259298e-06,
      "loss": 142.4745,
      "step": 21140
    },
    {
      "epoch": 0.1749596724159325,
      "grad_norm": 578.0482177734375,
      "learning_rate": 9.847615725553457e-06,
      "loss": 149.1546,
      "step": 21150
    },
    {
      "epoch": 0.17504239566530172,
      "grad_norm": 2596.45654296875,
      "learning_rate": 9.847269993145082e-06,
      "loss": 140.9354,
      "step": 21160
    },
    {
      "epoch": 0.17512511891467097,
      "grad_norm": 1056.8643798828125,
      "learning_rate": 9.84692387506168e-06,
      "loss": 138.6333,
      "step": 21170
    },
    {
      "epoch": 0.1752078421640402,
      "grad_norm": 543.871826171875,
      "learning_rate": 9.846577371330788e-06,
      "loss": 100.0174,
      "step": 21180
    },
    {
      "epoch": 0.17529056541340943,
      "grad_norm": 505.1315002441406,
      "learning_rate": 9.846230481979978e-06,
      "loss": 126.8892,
      "step": 21190
    },
    {
      "epoch": 0.17537328866277868,
      "grad_norm": 1152.1531982421875,
      "learning_rate": 9.84588320703685e-06,
      "loss": 137.2241,
      "step": 21200
    },
    {
      "epoch": 0.1754560119121479,
      "grad_norm": 2514.837646484375,
      "learning_rate": 9.845535546529036e-06,
      "loss": 150.1807,
      "step": 21210
    },
    {
      "epoch": 0.17553873516151713,
      "grad_norm": 1177.4676513671875,
      "learning_rate": 9.845187500484194e-06,
      "loss": 136.4538,
      "step": 21220
    },
    {
      "epoch": 0.1756214584108864,
      "grad_norm": 1041.1258544921875,
      "learning_rate": 9.844839068930021e-06,
      "loss": 131.5591,
      "step": 21230
    },
    {
      "epoch": 0.17570418166025562,
      "grad_norm": 907.1217651367188,
      "learning_rate": 9.844490251894237e-06,
      "loss": 129.923,
      "step": 21240
    },
    {
      "epoch": 0.17578690490962484,
      "grad_norm": 917.8026123046875,
      "learning_rate": 9.844141049404598e-06,
      "loss": 112.846,
      "step": 21250
    },
    {
      "epoch": 0.1758696281589941,
      "grad_norm": 1224.1368408203125,
      "learning_rate": 9.843791461488887e-06,
      "loss": 142.4482,
      "step": 21260
    },
    {
      "epoch": 0.17595235140836332,
      "grad_norm": 1229.176513671875,
      "learning_rate": 9.843441488174918e-06,
      "loss": 103.1861,
      "step": 21270
    },
    {
      "epoch": 0.17603507465773255,
      "grad_norm": 850.4046630859375,
      "learning_rate": 9.843091129490539e-06,
      "loss": 127.6695,
      "step": 21280
    },
    {
      "epoch": 0.1761177979071018,
      "grad_norm": 842.6237182617188,
      "learning_rate": 9.842740385463628e-06,
      "loss": 150.564,
      "step": 21290
    },
    {
      "epoch": 0.17620052115647103,
      "grad_norm": 1639.1712646484375,
      "learning_rate": 9.842389256122086e-06,
      "loss": 154.4747,
      "step": 21300
    },
    {
      "epoch": 0.17628324440584026,
      "grad_norm": 1027.3468017578125,
      "learning_rate": 9.842037741493856e-06,
      "loss": 138.2844,
      "step": 21310
    },
    {
      "epoch": 0.1763659676552095,
      "grad_norm": 1297.7607421875,
      "learning_rate": 9.841685841606905e-06,
      "loss": 171.9979,
      "step": 21320
    },
    {
      "epoch": 0.17644869090457874,
      "grad_norm": 650.8685302734375,
      "learning_rate": 9.841333556489232e-06,
      "loss": 138.841,
      "step": 21330
    },
    {
      "epoch": 0.17653141415394796,
      "grad_norm": 1357.2320556640625,
      "learning_rate": 9.840980886168866e-06,
      "loss": 151.9759,
      "step": 21340
    },
    {
      "epoch": 0.1766141374033172,
      "grad_norm": 938.8858032226562,
      "learning_rate": 9.840627830673867e-06,
      "loss": 141.0563,
      "step": 21350
    },
    {
      "epoch": 0.17669686065268644,
      "grad_norm": 1711.847412109375,
      "learning_rate": 9.84027439003233e-06,
      "loss": 172.4341,
      "step": 21360
    },
    {
      "epoch": 0.17677958390205567,
      "grad_norm": 854.195068359375,
      "learning_rate": 9.839920564272372e-06,
      "loss": 137.6896,
      "step": 21370
    },
    {
      "epoch": 0.1768623071514249,
      "grad_norm": 1087.6898193359375,
      "learning_rate": 9.839566353422148e-06,
      "loss": 117.6248,
      "step": 21380
    },
    {
      "epoch": 0.17694503040079415,
      "grad_norm": 2788.3251953125,
      "learning_rate": 9.839211757509838e-06,
      "loss": 144.8487,
      "step": 21390
    },
    {
      "epoch": 0.17702775365016338,
      "grad_norm": 1420.71875,
      "learning_rate": 9.83885677656366e-06,
      "loss": 122.336,
      "step": 21400
    },
    {
      "epoch": 0.1771104768995326,
      "grad_norm": 761.9530029296875,
      "learning_rate": 9.838501410611852e-06,
      "loss": 132.9662,
      "step": 21410
    },
    {
      "epoch": 0.17719320014890186,
      "grad_norm": 2090.32666015625,
      "learning_rate": 9.838145659682695e-06,
      "loss": 183.7629,
      "step": 21420
    },
    {
      "epoch": 0.17727592339827108,
      "grad_norm": 3058.696044921875,
      "learning_rate": 9.837789523804491e-06,
      "loss": 142.1503,
      "step": 21430
    },
    {
      "epoch": 0.1773586466476403,
      "grad_norm": 1059.2083740234375,
      "learning_rate": 9.837433003005578e-06,
      "loss": 137.1923,
      "step": 21440
    },
    {
      "epoch": 0.17744136989700957,
      "grad_norm": 1151.251708984375,
      "learning_rate": 9.83707609731432e-06,
      "loss": 89.3938,
      "step": 21450
    },
    {
      "epoch": 0.1775240931463788,
      "grad_norm": 749.4866333007812,
      "learning_rate": 9.836718806759119e-06,
      "loss": 145.9112,
      "step": 21460
    },
    {
      "epoch": 0.17760681639574802,
      "grad_norm": 1220.1741943359375,
      "learning_rate": 9.836361131368398e-06,
      "loss": 179.5797,
      "step": 21470
    },
    {
      "epoch": 0.17768953964511727,
      "grad_norm": 446.74859619140625,
      "learning_rate": 9.836003071170617e-06,
      "loss": 156.5806,
      "step": 21480
    },
    {
      "epoch": 0.1777722628944865,
      "grad_norm": 872.1563720703125,
      "learning_rate": 9.835644626194268e-06,
      "loss": 141.7276,
      "step": 21490
    },
    {
      "epoch": 0.17785498614385573,
      "grad_norm": 1689.736328125,
      "learning_rate": 9.835285796467867e-06,
      "loss": 133.7971,
      "step": 21500
    },
    {
      "epoch": 0.17793770939322495,
      "grad_norm": 1551.1575927734375,
      "learning_rate": 9.834926582019968e-06,
      "loss": 156.3386,
      "step": 21510
    },
    {
      "epoch": 0.1780204326425942,
      "grad_norm": 1045.95849609375,
      "learning_rate": 9.834566982879149e-06,
      "loss": 118.8335,
      "step": 21520
    },
    {
      "epoch": 0.17810315589196343,
      "grad_norm": 1034.3304443359375,
      "learning_rate": 9.83420699907402e-06,
      "loss": 136.8709,
      "step": 21530
    },
    {
      "epoch": 0.17818587914133266,
      "grad_norm": 870.5877075195312,
      "learning_rate": 9.83384663063323e-06,
      "loss": 125.3036,
      "step": 21540
    },
    {
      "epoch": 0.1782686023907019,
      "grad_norm": 3151.6611328125,
      "learning_rate": 9.833485877585447e-06,
      "loss": 230.3187,
      "step": 21550
    },
    {
      "epoch": 0.17835132564007114,
      "grad_norm": 1180.687255859375,
      "learning_rate": 9.833124739959375e-06,
      "loss": 128.1897,
      "step": 21560
    },
    {
      "epoch": 0.17843404888944037,
      "grad_norm": 1331.4376220703125,
      "learning_rate": 9.83276321778375e-06,
      "loss": 161.2605,
      "step": 21570
    },
    {
      "epoch": 0.17851677213880962,
      "grad_norm": 347.8481750488281,
      "learning_rate": 9.832401311087334e-06,
      "loss": 137.9425,
      "step": 21580
    },
    {
      "epoch": 0.17859949538817885,
      "grad_norm": 1399.0616455078125,
      "learning_rate": 9.832039019898922e-06,
      "loss": 141.6175,
      "step": 21590
    },
    {
      "epoch": 0.17868221863754807,
      "grad_norm": 1321.5462646484375,
      "learning_rate": 9.831676344247343e-06,
      "loss": 145.0367,
      "step": 21600
    },
    {
      "epoch": 0.17876494188691733,
      "grad_norm": 2181.245361328125,
      "learning_rate": 9.831313284161452e-06,
      "loss": 127.1829,
      "step": 21610
    },
    {
      "epoch": 0.17884766513628655,
      "grad_norm": 980.81201171875,
      "learning_rate": 9.830949839670134e-06,
      "loss": 160.9928,
      "step": 21620
    },
    {
      "epoch": 0.17893038838565578,
      "grad_norm": 1042.4224853515625,
      "learning_rate": 9.83058601080231e-06,
      "loss": 143.5391,
      "step": 21630
    },
    {
      "epoch": 0.17901311163502504,
      "grad_norm": 1248.8353271484375,
      "learning_rate": 9.830221797586925e-06,
      "loss": 125.7723,
      "step": 21640
    },
    {
      "epoch": 0.17909583488439426,
      "grad_norm": 765.4020385742188,
      "learning_rate": 9.829857200052961e-06,
      "loss": 145.4247,
      "step": 21650
    },
    {
      "epoch": 0.1791785581337635,
      "grad_norm": 1584.0721435546875,
      "learning_rate": 9.829492218229426e-06,
      "loss": 129.5774,
      "step": 21660
    },
    {
      "epoch": 0.17926128138313274,
      "grad_norm": 640.2371215820312,
      "learning_rate": 9.829126852145357e-06,
      "loss": 175.8682,
      "step": 21670
    },
    {
      "epoch": 0.17934400463250197,
      "grad_norm": 879.1619262695312,
      "learning_rate": 9.82876110182983e-06,
      "loss": 148.2943,
      "step": 21680
    },
    {
      "epoch": 0.1794267278818712,
      "grad_norm": 1808.1944580078125,
      "learning_rate": 9.82839496731194e-06,
      "loss": 161.7926,
      "step": 21690
    },
    {
      "epoch": 0.17950945113124042,
      "grad_norm": 1331.4281005859375,
      "learning_rate": 9.828028448620824e-06,
      "loss": 144.7912,
      "step": 21700
    },
    {
      "epoch": 0.17959217438060968,
      "grad_norm": 916.88134765625,
      "learning_rate": 9.827661545785641e-06,
      "loss": 111.8015,
      "step": 21710
    },
    {
      "epoch": 0.1796748976299789,
      "grad_norm": 945.744140625,
      "learning_rate": 9.827294258835584e-06,
      "loss": 135.7884,
      "step": 21720
    },
    {
      "epoch": 0.17975762087934813,
      "grad_norm": 843.4868774414062,
      "learning_rate": 9.82692658779988e-06,
      "loss": 166.0405,
      "step": 21730
    },
    {
      "epoch": 0.17984034412871738,
      "grad_norm": 866.8135375976562,
      "learning_rate": 9.826558532707777e-06,
      "loss": 131.8315,
      "step": 21740
    },
    {
      "epoch": 0.1799230673780866,
      "grad_norm": 1258.517822265625,
      "learning_rate": 9.826190093588564e-06,
      "loss": 192.2693,
      "step": 21750
    },
    {
      "epoch": 0.18000579062745584,
      "grad_norm": 755.1361694335938,
      "learning_rate": 9.825821270471555e-06,
      "loss": 110.0884,
      "step": 21760
    },
    {
      "epoch": 0.1800885138768251,
      "grad_norm": 505.46856689453125,
      "learning_rate": 9.825452063386094e-06,
      "loss": 117.8567,
      "step": 21770
    },
    {
      "epoch": 0.18017123712619432,
      "grad_norm": 1369.44287109375,
      "learning_rate": 9.825082472361558e-06,
      "loss": 135.8278,
      "step": 21780
    },
    {
      "epoch": 0.18025396037556354,
      "grad_norm": 917.852783203125,
      "learning_rate": 9.824712497427354e-06,
      "loss": 101.3896,
      "step": 21790
    },
    {
      "epoch": 0.1803366836249328,
      "grad_norm": 987.9921264648438,
      "learning_rate": 9.824342138612918e-06,
      "loss": 140.2833,
      "step": 21800
    },
    {
      "epoch": 0.18041940687430202,
      "grad_norm": 902.6442260742188,
      "learning_rate": 9.823971395947723e-06,
      "loss": 147.3234,
      "step": 21810
    },
    {
      "epoch": 0.18050213012367125,
      "grad_norm": 767.7576293945312,
      "learning_rate": 9.823600269461259e-06,
      "loss": 148.7868,
      "step": 21820
    },
    {
      "epoch": 0.1805848533730405,
      "grad_norm": 766.7611694335938,
      "learning_rate": 9.823228759183058e-06,
      "loss": 124.8973,
      "step": 21830
    },
    {
      "epoch": 0.18066757662240973,
      "grad_norm": 836.1171264648438,
      "learning_rate": 9.822856865142683e-06,
      "loss": 117.0898,
      "step": 21840
    },
    {
      "epoch": 0.18075029987177896,
      "grad_norm": 1208.75146484375,
      "learning_rate": 9.822484587369721e-06,
      "loss": 125.5648,
      "step": 21850
    },
    {
      "epoch": 0.1808330231211482,
      "grad_norm": 1158.348388671875,
      "learning_rate": 9.822111925893792e-06,
      "loss": 139.8755,
      "step": 21860
    },
    {
      "epoch": 0.18091574637051744,
      "grad_norm": 1267.1917724609375,
      "learning_rate": 9.821738880744549e-06,
      "loss": 135.5524,
      "step": 21870
    },
    {
      "epoch": 0.18099846961988666,
      "grad_norm": 2137.44580078125,
      "learning_rate": 9.82136545195167e-06,
      "loss": 159.7926,
      "step": 21880
    },
    {
      "epoch": 0.1810811928692559,
      "grad_norm": 955.8012084960938,
      "learning_rate": 9.82099163954487e-06,
      "loss": 121.1186,
      "step": 21890
    },
    {
      "epoch": 0.18116391611862515,
      "grad_norm": 1042.67724609375,
      "learning_rate": 9.820617443553889e-06,
      "loss": 132.7085,
      "step": 21900
    },
    {
      "epoch": 0.18124663936799437,
      "grad_norm": 988.7931518554688,
      "learning_rate": 9.820242864008503e-06,
      "loss": 141.3183,
      "step": 21910
    },
    {
      "epoch": 0.1813293626173636,
      "grad_norm": 1212.5155029296875,
      "learning_rate": 9.819867900938514e-06,
      "loss": 139.713,
      "step": 21920
    },
    {
      "epoch": 0.18141208586673285,
      "grad_norm": 714.6464233398438,
      "learning_rate": 9.819492554373758e-06,
      "loss": 122.3466,
      "step": 21930
    },
    {
      "epoch": 0.18149480911610208,
      "grad_norm": 1552.1221923828125,
      "learning_rate": 9.819116824344095e-06,
      "loss": 127.3137,
      "step": 21940
    },
    {
      "epoch": 0.1815775323654713,
      "grad_norm": 577.6259155273438,
      "learning_rate": 9.818740710879424e-06,
      "loss": 87.6241,
      "step": 21950
    },
    {
      "epoch": 0.18166025561484056,
      "grad_norm": 947.9591064453125,
      "learning_rate": 9.81836421400967e-06,
      "loss": 109.6651,
      "step": 21960
    },
    {
      "epoch": 0.1817429788642098,
      "grad_norm": 915.073486328125,
      "learning_rate": 9.81798733376479e-06,
      "loss": 112.4667,
      "step": 21970
    },
    {
      "epoch": 0.181825702113579,
      "grad_norm": 971.461181640625,
      "learning_rate": 9.817610070174768e-06,
      "loss": 159.3875,
      "step": 21980
    },
    {
      "epoch": 0.18190842536294827,
      "grad_norm": 839.3546142578125,
      "learning_rate": 9.817232423269622e-06,
      "loss": 141.2685,
      "step": 21990
    },
    {
      "epoch": 0.1819911486123175,
      "grad_norm": 901.919189453125,
      "learning_rate": 9.816854393079402e-06,
      "loss": 150.3302,
      "step": 22000
    },
    {
      "epoch": 0.18207387186168672,
      "grad_norm": 1099.916259765625,
      "learning_rate": 9.816475979634183e-06,
      "loss": 105.6784,
      "step": 22010
    },
    {
      "epoch": 0.18215659511105597,
      "grad_norm": 1539.3607177734375,
      "learning_rate": 9.816097182964076e-06,
      "loss": 150.3366,
      "step": 22020
    },
    {
      "epoch": 0.1822393183604252,
      "grad_norm": 1262.76513671875,
      "learning_rate": 9.81571800309922e-06,
      "loss": 205.2206,
      "step": 22030
    },
    {
      "epoch": 0.18232204160979443,
      "grad_norm": 1119.6072998046875,
      "learning_rate": 9.815338440069782e-06,
      "loss": 97.6272,
      "step": 22040
    },
    {
      "epoch": 0.18240476485916368,
      "grad_norm": 977.2727661132812,
      "learning_rate": 9.814958493905962e-06,
      "loss": 154.7452,
      "step": 22050
    },
    {
      "epoch": 0.1824874881085329,
      "grad_norm": 1467.2486572265625,
      "learning_rate": 9.814578164637996e-06,
      "loss": 116.3554,
      "step": 22060
    },
    {
      "epoch": 0.18257021135790213,
      "grad_norm": 1132.6796875,
      "learning_rate": 9.81419745229614e-06,
      "loss": 146.8583,
      "step": 22070
    },
    {
      "epoch": 0.18265293460727136,
      "grad_norm": 1301.7706298828125,
      "learning_rate": 9.813816356910685e-06,
      "loss": 150.6081,
      "step": 22080
    },
    {
      "epoch": 0.18273565785664062,
      "grad_norm": 786.4771728515625,
      "learning_rate": 9.813434878511956e-06,
      "loss": 98.5915,
      "step": 22090
    },
    {
      "epoch": 0.18281838110600984,
      "grad_norm": 1082.4107666015625,
      "learning_rate": 9.813053017130305e-06,
      "loss": 106.2249,
      "step": 22100
    },
    {
      "epoch": 0.18290110435537907,
      "grad_norm": 844.4359130859375,
      "learning_rate": 9.812670772796113e-06,
      "loss": 108.4317,
      "step": 22110
    },
    {
      "epoch": 0.18298382760474832,
      "grad_norm": 1356.7515869140625,
      "learning_rate": 9.812288145539796e-06,
      "loss": 156.1458,
      "step": 22120
    },
    {
      "epoch": 0.18306655085411755,
      "grad_norm": 571.9811401367188,
      "learning_rate": 9.811905135391796e-06,
      "loss": 128.8242,
      "step": 22130
    },
    {
      "epoch": 0.18314927410348678,
      "grad_norm": 996.9983520507812,
      "learning_rate": 9.81152174238259e-06,
      "loss": 121.8907,
      "step": 22140
    },
    {
      "epoch": 0.18323199735285603,
      "grad_norm": 1005.370361328125,
      "learning_rate": 9.81113796654268e-06,
      "loss": 130.7526,
      "step": 22150
    },
    {
      "epoch": 0.18331472060222526,
      "grad_norm": 522.5587158203125,
      "learning_rate": 9.810753807902603e-06,
      "loss": 167.8494,
      "step": 22160
    },
    {
      "epoch": 0.18339744385159448,
      "grad_norm": 1476.099853515625,
      "learning_rate": 9.81036926649292e-06,
      "loss": 125.2975,
      "step": 22170
    },
    {
      "epoch": 0.18348016710096374,
      "grad_norm": 864.9098510742188,
      "learning_rate": 9.809984342344234e-06,
      "loss": 114.3755,
      "step": 22180
    },
    {
      "epoch": 0.18356289035033296,
      "grad_norm": 1131.15966796875,
      "learning_rate": 9.80959903548717e-06,
      "loss": 130.3137,
      "step": 22190
    },
    {
      "epoch": 0.1836456135997022,
      "grad_norm": 2763.490234375,
      "learning_rate": 9.80921334595238e-06,
      "loss": 159.7871,
      "step": 22200
    },
    {
      "epoch": 0.18372833684907144,
      "grad_norm": 815.5933227539062,
      "learning_rate": 9.808827273770558e-06,
      "loss": 103.766,
      "step": 22210
    },
    {
      "epoch": 0.18381106009844067,
      "grad_norm": 1680.257568359375,
      "learning_rate": 9.80844081897242e-06,
      "loss": 162.1204,
      "step": 22220
    },
    {
      "epoch": 0.1838937833478099,
      "grad_norm": 934.7627563476562,
      "learning_rate": 9.808053981588712e-06,
      "loss": 144.453,
      "step": 22230
    },
    {
      "epoch": 0.18397650659717912,
      "grad_norm": 1144.36767578125,
      "learning_rate": 9.807666761650215e-06,
      "loss": 121.5437,
      "step": 22240
    },
    {
      "epoch": 0.18405922984654838,
      "grad_norm": 2383.975341796875,
      "learning_rate": 9.80727915918774e-06,
      "loss": 178.4958,
      "step": 22250
    },
    {
      "epoch": 0.1841419530959176,
      "grad_norm": 1073.094482421875,
      "learning_rate": 9.806891174232122e-06,
      "loss": 149.8745,
      "step": 22260
    },
    {
      "epoch": 0.18422467634528683,
      "grad_norm": 1233.2926025390625,
      "learning_rate": 9.806502806814236e-06,
      "loss": 128.9702,
      "step": 22270
    },
    {
      "epoch": 0.18430739959465609,
      "grad_norm": 837.9373168945312,
      "learning_rate": 9.806114056964977e-06,
      "loss": 139.9306,
      "step": 22280
    },
    {
      "epoch": 0.1843901228440253,
      "grad_norm": 628.1285400390625,
      "learning_rate": 9.805724924715283e-06,
      "loss": 123.0449,
      "step": 22290
    },
    {
      "epoch": 0.18447284609339454,
      "grad_norm": 561.8240356445312,
      "learning_rate": 9.80533541009611e-06,
      "loss": 105.3535,
      "step": 22300
    },
    {
      "epoch": 0.1845555693427638,
      "grad_norm": 1421.851806640625,
      "learning_rate": 9.804945513138454e-06,
      "loss": 219.4902,
      "step": 22310
    },
    {
      "epoch": 0.18463829259213302,
      "grad_norm": 825.07080078125,
      "learning_rate": 9.804555233873335e-06,
      "loss": 135.7106,
      "step": 22320
    },
    {
      "epoch": 0.18472101584150225,
      "grad_norm": 2163.25439453125,
      "learning_rate": 9.804164572331804e-06,
      "loss": 173.7582,
      "step": 22330
    },
    {
      "epoch": 0.1848037390908715,
      "grad_norm": 1122.6065673828125,
      "learning_rate": 9.80377352854495e-06,
      "loss": 91.7703,
      "step": 22340
    },
    {
      "epoch": 0.18488646234024073,
      "grad_norm": 909.8526000976562,
      "learning_rate": 9.80338210254388e-06,
      "loss": 124.849,
      "step": 22350
    },
    {
      "epoch": 0.18496918558960995,
      "grad_norm": 784.2378540039062,
      "learning_rate": 9.80299029435974e-06,
      "loss": 148.4874,
      "step": 22360
    },
    {
      "epoch": 0.1850519088389792,
      "grad_norm": 999.297119140625,
      "learning_rate": 9.802598104023706e-06,
      "loss": 159.1145,
      "step": 22370
    },
    {
      "epoch": 0.18513463208834843,
      "grad_norm": 753.5262451171875,
      "learning_rate": 9.80220553156698e-06,
      "loss": 113.2028,
      "step": 22380
    },
    {
      "epoch": 0.18521735533771766,
      "grad_norm": 1254.95947265625,
      "learning_rate": 9.801812577020802e-06,
      "loss": 139.2906,
      "step": 22390
    },
    {
      "epoch": 0.1853000785870869,
      "grad_norm": 1023.6962890625,
      "learning_rate": 9.801419240416432e-06,
      "loss": 144.157,
      "step": 22400
    },
    {
      "epoch": 0.18538280183645614,
      "grad_norm": 937.4168701171875,
      "learning_rate": 9.80102552178517e-06,
      "loss": 124.0214,
      "step": 22410
    },
    {
      "epoch": 0.18546552508582537,
      "grad_norm": 1038.947265625,
      "learning_rate": 9.800631421158341e-06,
      "loss": 134.8847,
      "step": 22420
    },
    {
      "epoch": 0.1855482483351946,
      "grad_norm": 839.5892333984375,
      "learning_rate": 9.800236938567302e-06,
      "loss": 114.9243,
      "step": 22430
    },
    {
      "epoch": 0.18563097158456385,
      "grad_norm": 1064.1595458984375,
      "learning_rate": 9.799842074043438e-06,
      "loss": 132.5262,
      "step": 22440
    },
    {
      "epoch": 0.18571369483393307,
      "grad_norm": 868.5515747070312,
      "learning_rate": 9.799446827618172e-06,
      "loss": 123.4177,
      "step": 22450
    },
    {
      "epoch": 0.1857964180833023,
      "grad_norm": 1418.9991455078125,
      "learning_rate": 9.799051199322944e-06,
      "loss": 134.293,
      "step": 22460
    },
    {
      "epoch": 0.18587914133267155,
      "grad_norm": 1264.387939453125,
      "learning_rate": 9.798655189189239e-06,
      "loss": 155.6345,
      "step": 22470
    },
    {
      "epoch": 0.18596186458204078,
      "grad_norm": 1341.15185546875,
      "learning_rate": 9.798258797248563e-06,
      "loss": 151.904,
      "step": 22480
    },
    {
      "epoch": 0.18604458783141,
      "grad_norm": 1191.142578125,
      "learning_rate": 9.797862023532457e-06,
      "loss": 143.8828,
      "step": 22490
    },
    {
      "epoch": 0.18612731108077926,
      "grad_norm": 1029.2672119140625,
      "learning_rate": 9.797464868072489e-06,
      "loss": 111.6996,
      "step": 22500
    },
    {
      "epoch": 0.1862100343301485,
      "grad_norm": 898.6533203125,
      "learning_rate": 9.797067330900256e-06,
      "loss": 123.9605,
      "step": 22510
    },
    {
      "epoch": 0.18629275757951771,
      "grad_norm": 1148.5517578125,
      "learning_rate": 9.796669412047392e-06,
      "loss": 107.9766,
      "step": 22520
    },
    {
      "epoch": 0.18637548082888697,
      "grad_norm": 1347.3740234375,
      "learning_rate": 9.796271111545559e-06,
      "loss": 132.5673,
      "step": 22530
    },
    {
      "epoch": 0.1864582040782562,
      "grad_norm": 732.031982421875,
      "learning_rate": 9.795872429426443e-06,
      "loss": 113.1994,
      "step": 22540
    },
    {
      "epoch": 0.18654092732762542,
      "grad_norm": 847.0791625976562,
      "learning_rate": 9.79547336572177e-06,
      "loss": 115.746,
      "step": 22550
    },
    {
      "epoch": 0.18662365057699468,
      "grad_norm": 897.2965698242188,
      "learning_rate": 9.795073920463289e-06,
      "loss": 123.8876,
      "step": 22560
    },
    {
      "epoch": 0.1867063738263639,
      "grad_norm": 1148.21875,
      "learning_rate": 9.794674093682781e-06,
      "loss": 133.7378,
      "step": 22570
    },
    {
      "epoch": 0.18678909707573313,
      "grad_norm": 1180.834716796875,
      "learning_rate": 9.79427388541206e-06,
      "loss": 100.9495,
      "step": 22580
    },
    {
      "epoch": 0.18687182032510238,
      "grad_norm": 681.8207397460938,
      "learning_rate": 9.79387329568297e-06,
      "loss": 108.7203,
      "step": 22590
    },
    {
      "epoch": 0.1869545435744716,
      "grad_norm": 859.6419067382812,
      "learning_rate": 9.793472324527383e-06,
      "loss": 179.532,
      "step": 22600
    },
    {
      "epoch": 0.18703726682384084,
      "grad_norm": 615.40380859375,
      "learning_rate": 9.793070971977203e-06,
      "loss": 103.2244,
      "step": 22610
    },
    {
      "epoch": 0.18711999007321006,
      "grad_norm": 896.5668334960938,
      "learning_rate": 9.79266923806436e-06,
      "loss": 135.8797,
      "step": 22620
    },
    {
      "epoch": 0.18720271332257932,
      "grad_norm": 685.52685546875,
      "learning_rate": 9.792267122820823e-06,
      "loss": 102.7591,
      "step": 22630
    },
    {
      "epoch": 0.18728543657194854,
      "grad_norm": 656.8187255859375,
      "learning_rate": 9.791864626278584e-06,
      "loss": 134.8737,
      "step": 22640
    },
    {
      "epoch": 0.18736815982131777,
      "grad_norm": 406.96820068359375,
      "learning_rate": 9.791461748469669e-06,
      "loss": 129.8552,
      "step": 22650
    },
    {
      "epoch": 0.18745088307068702,
      "grad_norm": 1129.7803955078125,
      "learning_rate": 9.791058489426134e-06,
      "loss": 133.4618,
      "step": 22660
    },
    {
      "epoch": 0.18753360632005625,
      "grad_norm": 1034.3385009765625,
      "learning_rate": 9.790654849180059e-06,
      "loss": 107.0448,
      "step": 22670
    },
    {
      "epoch": 0.18761632956942548,
      "grad_norm": 1405.0714111328125,
      "learning_rate": 9.790250827763566e-06,
      "loss": 121.3317,
      "step": 22680
    },
    {
      "epoch": 0.18769905281879473,
      "grad_norm": 1143.3116455078125,
      "learning_rate": 9.7898464252088e-06,
      "loss": 93.2796,
      "step": 22690
    },
    {
      "epoch": 0.18778177606816396,
      "grad_norm": 1161.796630859375,
      "learning_rate": 9.789441641547935e-06,
      "loss": 112.5664,
      "step": 22700
    },
    {
      "epoch": 0.18786449931753318,
      "grad_norm": 1665.2811279296875,
      "learning_rate": 9.789036476813178e-06,
      "loss": 162.8835,
      "step": 22710
    },
    {
      "epoch": 0.18794722256690244,
      "grad_norm": 885.5316772460938,
      "learning_rate": 9.788630931036769e-06,
      "loss": 127.6729,
      "step": 22720
    },
    {
      "epoch": 0.18802994581627167,
      "grad_norm": 1863.15673828125,
      "learning_rate": 9.788225004250974e-06,
      "loss": 136.1164,
      "step": 22730
    },
    {
      "epoch": 0.1881126690656409,
      "grad_norm": 770.5360717773438,
      "learning_rate": 9.78781869648809e-06,
      "loss": 145.4966,
      "step": 22740
    },
    {
      "epoch": 0.18819539231501015,
      "grad_norm": 1846.6016845703125,
      "learning_rate": 9.787412007780445e-06,
      "loss": 119.8534,
      "step": 22750
    },
    {
      "epoch": 0.18827811556437937,
      "grad_norm": 930.654052734375,
      "learning_rate": 9.787004938160398e-06,
      "loss": 112.2845,
      "step": 22760
    },
    {
      "epoch": 0.1883608388137486,
      "grad_norm": 3968.6298828125,
      "learning_rate": 9.786597487660336e-06,
      "loss": 166.1542,
      "step": 22770
    },
    {
      "epoch": 0.18844356206311783,
      "grad_norm": 687.5050659179688,
      "learning_rate": 9.78618965631268e-06,
      "loss": 118.5364,
      "step": 22780
    },
    {
      "epoch": 0.18852628531248708,
      "grad_norm": 1421.8592529296875,
      "learning_rate": 9.785781444149883e-06,
      "loss": 102.7248,
      "step": 22790
    },
    {
      "epoch": 0.1886090085618563,
      "grad_norm": 1723.80029296875,
      "learning_rate": 9.785372851204415e-06,
      "loss": 136.1481,
      "step": 22800
    },
    {
      "epoch": 0.18869173181122553,
      "grad_norm": 1364.1827392578125,
      "learning_rate": 9.784963877508794e-06,
      "loss": 135.9108,
      "step": 22810
    },
    {
      "epoch": 0.1887744550605948,
      "grad_norm": 2061.134765625,
      "learning_rate": 9.784554523095554e-06,
      "loss": 201.537,
      "step": 22820
    },
    {
      "epoch": 0.188857178309964,
      "grad_norm": 958.0075073242188,
      "learning_rate": 9.784144787997272e-06,
      "loss": 96.8345,
      "step": 22830
    },
    {
      "epoch": 0.18893990155933324,
      "grad_norm": 717.226318359375,
      "learning_rate": 9.783734672246545e-06,
      "loss": 130.6322,
      "step": 22840
    },
    {
      "epoch": 0.1890226248087025,
      "grad_norm": 716.5567016601562,
      "learning_rate": 9.783324175876004e-06,
      "loss": 105.1676,
      "step": 22850
    },
    {
      "epoch": 0.18910534805807172,
      "grad_norm": 1025.28173828125,
      "learning_rate": 9.782913298918311e-06,
      "loss": 151.713,
      "step": 22860
    },
    {
      "epoch": 0.18918807130744095,
      "grad_norm": 649.5341186523438,
      "learning_rate": 9.782502041406157e-06,
      "loss": 135.4802,
      "step": 22870
    },
    {
      "epoch": 0.1892707945568102,
      "grad_norm": 1117.214599609375,
      "learning_rate": 9.782090403372263e-06,
      "loss": 148.4325,
      "step": 22880
    },
    {
      "epoch": 0.18935351780617943,
      "grad_norm": 1014.591552734375,
      "learning_rate": 9.781678384849385e-06,
      "loss": 111.5626,
      "step": 22890
    },
    {
      "epoch": 0.18943624105554865,
      "grad_norm": 2505.06982421875,
      "learning_rate": 9.7812659858703e-06,
      "loss": 142.1111,
      "step": 22900
    },
    {
      "epoch": 0.1895189643049179,
      "grad_norm": 1313.8260498046875,
      "learning_rate": 9.780853206467826e-06,
      "loss": 133.5671,
      "step": 22910
    },
    {
      "epoch": 0.18960168755428713,
      "grad_norm": 984.4515380859375,
      "learning_rate": 9.780440046674803e-06,
      "loss": 167.4173,
      "step": 22920
    },
    {
      "epoch": 0.18968441080365636,
      "grad_norm": 893.4295043945312,
      "learning_rate": 9.780026506524106e-06,
      "loss": 159.9201,
      "step": 22930
    },
    {
      "epoch": 0.18976713405302562,
      "grad_norm": 818.2760620117188,
      "learning_rate": 9.779612586048635e-06,
      "loss": 132.297,
      "step": 22940
    },
    {
      "epoch": 0.18984985730239484,
      "grad_norm": 1028.792236328125,
      "learning_rate": 9.779198285281326e-06,
      "loss": 175.7389,
      "step": 22950
    },
    {
      "epoch": 0.18993258055176407,
      "grad_norm": 1137.0174560546875,
      "learning_rate": 9.778783604255145e-06,
      "loss": 120.6092,
      "step": 22960
    },
    {
      "epoch": 0.1900153038011333,
      "grad_norm": 2498.3369140625,
      "learning_rate": 9.778368543003083e-06,
      "loss": 144.8033,
      "step": 22970
    },
    {
      "epoch": 0.19009802705050255,
      "grad_norm": 1361.688232421875,
      "learning_rate": 9.777953101558164e-06,
      "loss": 133.7076,
      "step": 22980
    },
    {
      "epoch": 0.19018075029987178,
      "grad_norm": 1047.64453125,
      "learning_rate": 9.777537279953448e-06,
      "loss": 120.5423,
      "step": 22990
    },
    {
      "epoch": 0.190263473549241,
      "grad_norm": 982.462158203125,
      "learning_rate": 9.777121078222015e-06,
      "loss": 108.502,
      "step": 23000
    },
    {
      "epoch": 0.19034619679861026,
      "grad_norm": 1133.6990966796875,
      "learning_rate": 9.77670449639698e-06,
      "loss": 129.8804,
      "step": 23010
    },
    {
      "epoch": 0.19042892004797948,
      "grad_norm": 671.9098510742188,
      "learning_rate": 9.776287534511492e-06,
      "loss": 130.2723,
      "step": 23020
    },
    {
      "epoch": 0.1905116432973487,
      "grad_norm": 1220.64892578125,
      "learning_rate": 9.775870192598726e-06,
      "loss": 122.8446,
      "step": 23030
    },
    {
      "epoch": 0.19059436654671796,
      "grad_norm": 2241.37548828125,
      "learning_rate": 9.775452470691886e-06,
      "loss": 133.7913,
      "step": 23040
    },
    {
      "epoch": 0.1906770897960872,
      "grad_norm": 1148.03662109375,
      "learning_rate": 9.77503436882421e-06,
      "loss": 138.1318,
      "step": 23050
    },
    {
      "epoch": 0.19075981304545642,
      "grad_norm": 980.452880859375,
      "learning_rate": 9.774615887028964e-06,
      "loss": 147.464,
      "step": 23060
    },
    {
      "epoch": 0.19084253629482567,
      "grad_norm": 2212.347412109375,
      "learning_rate": 9.774197025339442e-06,
      "loss": 102.5226,
      "step": 23070
    },
    {
      "epoch": 0.1909252595441949,
      "grad_norm": 928.7798461914062,
      "learning_rate": 9.773777783788976e-06,
      "loss": 113.3667,
      "step": 23080
    },
    {
      "epoch": 0.19100798279356412,
      "grad_norm": 1282.0994873046875,
      "learning_rate": 9.77335816241092e-06,
      "loss": 144.8801,
      "step": 23090
    },
    {
      "epoch": 0.19109070604293338,
      "grad_norm": 837.4090576171875,
      "learning_rate": 9.77293816123866e-06,
      "loss": 104.0558,
      "step": 23100
    },
    {
      "epoch": 0.1911734292923026,
      "grad_norm": 881.4822998046875,
      "learning_rate": 9.772517780305618e-06,
      "loss": 121.4086,
      "step": 23110
    },
    {
      "epoch": 0.19125615254167183,
      "grad_norm": 880.7931518554688,
      "learning_rate": 9.772097019645236e-06,
      "loss": 139.5056,
      "step": 23120
    },
    {
      "epoch": 0.19133887579104109,
      "grad_norm": 776.1262817382812,
      "learning_rate": 9.771675879290998e-06,
      "loss": 146.6072,
      "step": 23130
    },
    {
      "epoch": 0.1914215990404103,
      "grad_norm": 904.765380859375,
      "learning_rate": 9.771254359276407e-06,
      "loss": 116.9324,
      "step": 23140
    },
    {
      "epoch": 0.19150432228977954,
      "grad_norm": 730.8990478515625,
      "learning_rate": 9.770832459635004e-06,
      "loss": 133.2764,
      "step": 23150
    },
    {
      "epoch": 0.19158704553914876,
      "grad_norm": 690.1975708007812,
      "learning_rate": 9.77041018040036e-06,
      "loss": 175.1648,
      "step": 23160
    },
    {
      "epoch": 0.19166976878851802,
      "grad_norm": 907.7679443359375,
      "learning_rate": 9.769987521606068e-06,
      "loss": 105.3016,
      "step": 23170
    },
    {
      "epoch": 0.19175249203788725,
      "grad_norm": 693.336181640625,
      "learning_rate": 9.769564483285761e-06,
      "loss": 192.6149,
      "step": 23180
    },
    {
      "epoch": 0.19183521528725647,
      "grad_norm": 557.3772583007812,
      "learning_rate": 9.769141065473099e-06,
      "loss": 115.1971,
      "step": 23190
    },
    {
      "epoch": 0.19191793853662573,
      "grad_norm": 858.6890869140625,
      "learning_rate": 9.768717268201768e-06,
      "loss": 145.7314,
      "step": 23200
    },
    {
      "epoch": 0.19200066178599495,
      "grad_norm": 2752.160888671875,
      "learning_rate": 9.768293091505491e-06,
      "loss": 128.7382,
      "step": 23210
    },
    {
      "epoch": 0.19208338503536418,
      "grad_norm": 971.1498413085938,
      "learning_rate": 9.767868535418014e-06,
      "loss": 126.5057,
      "step": 23220
    },
    {
      "epoch": 0.19216610828473343,
      "grad_norm": 793.6380615234375,
      "learning_rate": 9.767443599973122e-06,
      "loss": 108.5758,
      "step": 23230
    },
    {
      "epoch": 0.19224883153410266,
      "grad_norm": 1140.6827392578125,
      "learning_rate": 9.76701828520462e-06,
      "loss": 117.5244,
      "step": 23240
    },
    {
      "epoch": 0.1923315547834719,
      "grad_norm": 840.5177001953125,
      "learning_rate": 9.766592591146353e-06,
      "loss": 107.5938,
      "step": 23250
    },
    {
      "epoch": 0.19241427803284114,
      "grad_norm": 1344.9217529296875,
      "learning_rate": 9.766166517832188e-06,
      "loss": 120.1126,
      "step": 23260
    },
    {
      "epoch": 0.19249700128221037,
      "grad_norm": 1015.5147705078125,
      "learning_rate": 9.765740065296025e-06,
      "loss": 131.5013,
      "step": 23270
    },
    {
      "epoch": 0.1925797245315796,
      "grad_norm": 574.64306640625,
      "learning_rate": 9.765313233571798e-06,
      "loss": 131.5927,
      "step": 23280
    },
    {
      "epoch": 0.19266244778094885,
      "grad_norm": 1062.6119384765625,
      "learning_rate": 9.76488602269347e-06,
      "loss": 162.6192,
      "step": 23290
    },
    {
      "epoch": 0.19274517103031807,
      "grad_norm": 1031.8822021484375,
      "learning_rate": 9.764458432695026e-06,
      "loss": 91.6339,
      "step": 23300
    },
    {
      "epoch": 0.1928278942796873,
      "grad_norm": 715.5499877929688,
      "learning_rate": 9.76403046361049e-06,
      "loss": 98.5823,
      "step": 23310
    },
    {
      "epoch": 0.19291061752905656,
      "grad_norm": 707.9616088867188,
      "learning_rate": 9.763602115473914e-06,
      "loss": 129.4643,
      "step": 23320
    },
    {
      "epoch": 0.19299334077842578,
      "grad_norm": 1023.2615966796875,
      "learning_rate": 9.763173388319381e-06,
      "loss": 108.6447,
      "step": 23330
    },
    {
      "epoch": 0.193076064027795,
      "grad_norm": 745.7240600585938,
      "learning_rate": 9.762744282181e-06,
      "loss": 160.1926,
      "step": 23340
    },
    {
      "epoch": 0.19315878727716423,
      "grad_norm": 1169.117431640625,
      "learning_rate": 9.762314797092916e-06,
      "loss": 136.8101,
      "step": 23350
    },
    {
      "epoch": 0.1932415105265335,
      "grad_norm": 2356.3876953125,
      "learning_rate": 9.761884933089301e-06,
      "loss": 149.6885,
      "step": 23360
    },
    {
      "epoch": 0.19332423377590272,
      "grad_norm": 748.1530151367188,
      "learning_rate": 9.761454690204352e-06,
      "loss": 90.2527,
      "step": 23370
    },
    {
      "epoch": 0.19340695702527194,
      "grad_norm": 773.795654296875,
      "learning_rate": 9.76102406847231e-06,
      "loss": 133.7801,
      "step": 23380
    },
    {
      "epoch": 0.1934896802746412,
      "grad_norm": 785.1251831054688,
      "learning_rate": 9.760593067927428e-06,
      "loss": 92.1381,
      "step": 23390
    },
    {
      "epoch": 0.19357240352401042,
      "grad_norm": 987.9070434570312,
      "learning_rate": 9.760161688604008e-06,
      "loss": 152.8994,
      "step": 23400
    },
    {
      "epoch": 0.19365512677337965,
      "grad_norm": 1091.9166259765625,
      "learning_rate": 9.759729930536367e-06,
      "loss": 106.4156,
      "step": 23410
    },
    {
      "epoch": 0.1937378500227489,
      "grad_norm": 828.226806640625,
      "learning_rate": 9.75929779375886e-06,
      "loss": 127.5234,
      "step": 23420
    },
    {
      "epoch": 0.19382057327211813,
      "grad_norm": 1164.93359375,
      "learning_rate": 9.75886527830587e-06,
      "loss": 147.2189,
      "step": 23430
    },
    {
      "epoch": 0.19390329652148736,
      "grad_norm": 1556.861572265625,
      "learning_rate": 9.75843238421181e-06,
      "loss": 132.9836,
      "step": 23440
    },
    {
      "epoch": 0.1939860197708566,
      "grad_norm": 1154.7764892578125,
      "learning_rate": 9.757999111511121e-06,
      "loss": 131.7635,
      "step": 23450
    },
    {
      "epoch": 0.19406874302022584,
      "grad_norm": 1170.080078125,
      "learning_rate": 9.757565460238281e-06,
      "loss": 120.619,
      "step": 23460
    },
    {
      "epoch": 0.19415146626959506,
      "grad_norm": 1463.3924560546875,
      "learning_rate": 9.757131430427791e-06,
      "loss": 129.6818,
      "step": 23470
    },
    {
      "epoch": 0.19423418951896432,
      "grad_norm": 1100.9061279296875,
      "learning_rate": 9.756697022114185e-06,
      "loss": 108.3849,
      "step": 23480
    },
    {
      "epoch": 0.19431691276833354,
      "grad_norm": 1876.0504150390625,
      "learning_rate": 9.756262235332029e-06,
      "loss": 103.7323,
      "step": 23490
    },
    {
      "epoch": 0.19439963601770277,
      "grad_norm": 974.3872680664062,
      "learning_rate": 9.755827070115915e-06,
      "loss": 105.9175,
      "step": 23500
    },
    {
      "epoch": 0.194482359267072,
      "grad_norm": 888.02099609375,
      "learning_rate": 9.755391526500466e-06,
      "loss": 108.9482,
      "step": 23510
    },
    {
      "epoch": 0.19456508251644125,
      "grad_norm": 863.0517578125,
      "learning_rate": 9.75495560452034e-06,
      "loss": 131.4029,
      "step": 23520
    },
    {
      "epoch": 0.19464780576581048,
      "grad_norm": 1131.92431640625,
      "learning_rate": 9.754519304210214e-06,
      "loss": 122.4951,
      "step": 23530
    },
    {
      "epoch": 0.1947305290151797,
      "grad_norm": 2271.8134765625,
      "learning_rate": 9.754082625604812e-06,
      "loss": 121.6546,
      "step": 23540
    },
    {
      "epoch": 0.19481325226454896,
      "grad_norm": 919.2047119140625,
      "learning_rate": 9.753645568738872e-06,
      "loss": 138.8903,
      "step": 23550
    },
    {
      "epoch": 0.19489597551391818,
      "grad_norm": 840.5191650390625,
      "learning_rate": 9.75320813364717e-06,
      "loss": 91.0436,
      "step": 23560
    },
    {
      "epoch": 0.1949786987632874,
      "grad_norm": 621.6953125,
      "learning_rate": 9.752770320364512e-06,
      "loss": 128.6445,
      "step": 23570
    },
    {
      "epoch": 0.19506142201265667,
      "grad_norm": 754.2584838867188,
      "learning_rate": 9.752332128925732e-06,
      "loss": 111.4495,
      "step": 23580
    },
    {
      "epoch": 0.1951441452620259,
      "grad_norm": 1024.14501953125,
      "learning_rate": 9.751893559365693e-06,
      "loss": 170.7815,
      "step": 23590
    },
    {
      "epoch": 0.19522686851139512,
      "grad_norm": 1205.4876708984375,
      "learning_rate": 9.751454611719294e-06,
      "loss": 121.6351,
      "step": 23600
    },
    {
      "epoch": 0.19530959176076437,
      "grad_norm": 1503.17236328125,
      "learning_rate": 9.751015286021455e-06,
      "loss": 164.3172,
      "step": 23610
    },
    {
      "epoch": 0.1953923150101336,
      "grad_norm": 937.5809936523438,
      "learning_rate": 9.750575582307136e-06,
      "loss": 160.1191,
      "step": 23620
    },
    {
      "epoch": 0.19547503825950283,
      "grad_norm": 983.7006225585938,
      "learning_rate": 9.75013550061132e-06,
      "loss": 120.9869,
      "step": 23630
    },
    {
      "epoch": 0.19555776150887208,
      "grad_norm": 1772.2957763671875,
      "learning_rate": 9.749695040969022e-06,
      "loss": 114.3774,
      "step": 23640
    },
    {
      "epoch": 0.1956404847582413,
      "grad_norm": 924.7174072265625,
      "learning_rate": 9.749254203415288e-06,
      "loss": 124.1198,
      "step": 23650
    },
    {
      "epoch": 0.19572320800761053,
      "grad_norm": 972.820068359375,
      "learning_rate": 9.748812987985193e-06,
      "loss": 123.3391,
      "step": 23660
    },
    {
      "epoch": 0.1958059312569798,
      "grad_norm": 1330.291748046875,
      "learning_rate": 9.748371394713842e-06,
      "loss": 126.0948,
      "step": 23670
    },
    {
      "epoch": 0.195888654506349,
      "grad_norm": 1056.2666015625,
      "learning_rate": 9.747929423636372e-06,
      "loss": 128.4867,
      "step": 23680
    },
    {
      "epoch": 0.19597137775571824,
      "grad_norm": 670.9091186523438,
      "learning_rate": 9.74748707478795e-06,
      "loss": 140.6503,
      "step": 23690
    },
    {
      "epoch": 0.19605410100508747,
      "grad_norm": 1226.66162109375,
      "learning_rate": 9.747044348203766e-06,
      "loss": 130.9208,
      "step": 23700
    },
    {
      "epoch": 0.19613682425445672,
      "grad_norm": 907.8272705078125,
      "learning_rate": 9.74660124391905e-06,
      "loss": 127.5544,
      "step": 23710
    },
    {
      "epoch": 0.19621954750382595,
      "grad_norm": 898.1268920898438,
      "learning_rate": 9.746157761969058e-06,
      "loss": 164.2529,
      "step": 23720
    },
    {
      "epoch": 0.19630227075319517,
      "grad_norm": 1068.112060546875,
      "learning_rate": 9.745713902389074e-06,
      "loss": 143.5591,
      "step": 23730
    },
    {
      "epoch": 0.19638499400256443,
      "grad_norm": 1102.2633056640625,
      "learning_rate": 9.745269665214415e-06,
      "loss": 122.1263,
      "step": 23740
    },
    {
      "epoch": 0.19646771725193365,
      "grad_norm": 556.4323120117188,
      "learning_rate": 9.744825050480425e-06,
      "loss": 121.492,
      "step": 23750
    },
    {
      "epoch": 0.19655044050130288,
      "grad_norm": 1045.6256103515625,
      "learning_rate": 9.744380058222483e-06,
      "loss": 131.9083,
      "step": 23760
    },
    {
      "epoch": 0.19663316375067214,
      "grad_norm": 1136.8719482421875,
      "learning_rate": 9.743934688475994e-06,
      "loss": 115.8493,
      "step": 23770
    },
    {
      "epoch": 0.19671588700004136,
      "grad_norm": 962.9896850585938,
      "learning_rate": 9.743488941276394e-06,
      "loss": 117.902,
      "step": 23780
    },
    {
      "epoch": 0.1967986102494106,
      "grad_norm": 1086.8525390625,
      "learning_rate": 9.743042816659147e-06,
      "loss": 134.2704,
      "step": 23790
    },
    {
      "epoch": 0.19688133349877984,
      "grad_norm": 1227.1202392578125,
      "learning_rate": 9.742596314659751e-06,
      "loss": 127.5619,
      "step": 23800
    },
    {
      "epoch": 0.19696405674814907,
      "grad_norm": 1608.360595703125,
      "learning_rate": 9.742149435313732e-06,
      "loss": 120.5991,
      "step": 23810
    },
    {
      "epoch": 0.1970467799975183,
      "grad_norm": 664.271728515625,
      "learning_rate": 9.741702178656647e-06,
      "loss": 151.2022,
      "step": 23820
    },
    {
      "epoch": 0.19712950324688755,
      "grad_norm": 1345.0858154296875,
      "learning_rate": 9.74125454472408e-06,
      "loss": 136.8234,
      "step": 23830
    },
    {
      "epoch": 0.19721222649625678,
      "grad_norm": 732.8429565429688,
      "learning_rate": 9.740806533551647e-06,
      "loss": 234.7962,
      "step": 23840
    },
    {
      "epoch": 0.197294949745626,
      "grad_norm": 1542.8756103515625,
      "learning_rate": 9.740358145174999e-06,
      "loss": 130.3346,
      "step": 23850
    },
    {
      "epoch": 0.19737767299499526,
      "grad_norm": 821.8853149414062,
      "learning_rate": 9.739909379629805e-06,
      "loss": 139.0029,
      "step": 23860
    },
    {
      "epoch": 0.19746039624436448,
      "grad_norm": 812.029541015625,
      "learning_rate": 9.739460236951778e-06,
      "loss": 127.8701,
      "step": 23870
    },
    {
      "epoch": 0.1975431194937337,
      "grad_norm": 969.9833984375,
      "learning_rate": 9.739010717176649e-06,
      "loss": 99.5856,
      "step": 23880
    },
    {
      "epoch": 0.19762584274310294,
      "grad_norm": 1117.0540771484375,
      "learning_rate": 9.738560820340189e-06,
      "loss": 114.7225,
      "step": 23890
    },
    {
      "epoch": 0.1977085659924722,
      "grad_norm": 601.694091796875,
      "learning_rate": 9.738110546478188e-06,
      "loss": 123.2165,
      "step": 23900
    },
    {
      "epoch": 0.19779128924184142,
      "grad_norm": 1268.921630859375,
      "learning_rate": 9.737659895626478e-06,
      "loss": 130.8849,
      "step": 23910
    },
    {
      "epoch": 0.19787401249121064,
      "grad_norm": 1429.4453125,
      "learning_rate": 9.737208867820914e-06,
      "loss": 123.7755,
      "step": 23920
    },
    {
      "epoch": 0.1979567357405799,
      "grad_norm": 654.82080078125,
      "learning_rate": 9.736757463097378e-06,
      "loss": 152.5086,
      "step": 23930
    },
    {
      "epoch": 0.19803945898994912,
      "grad_norm": 852.7525024414062,
      "learning_rate": 9.736305681491792e-06,
      "loss": 116.6475,
      "step": 23940
    },
    {
      "epoch": 0.19812218223931835,
      "grad_norm": 1039.8662109375,
      "learning_rate": 9.735853523040098e-06,
      "loss": 138.9281,
      "step": 23950
    },
    {
      "epoch": 0.1982049054886876,
      "grad_norm": 1178.4395751953125,
      "learning_rate": 9.735400987778274e-06,
      "loss": 143.172,
      "step": 23960
    },
    {
      "epoch": 0.19828762873805683,
      "grad_norm": 1232.4547119140625,
      "learning_rate": 9.734948075742328e-06,
      "loss": 121.5333,
      "step": 23970
    },
    {
      "epoch": 0.19837035198742606,
      "grad_norm": 865.7545166015625,
      "learning_rate": 9.734494786968293e-06,
      "loss": 154.3453,
      "step": 23980
    },
    {
      "epoch": 0.1984530752367953,
      "grad_norm": 669.6997680664062,
      "learning_rate": 9.734041121492235e-06,
      "loss": 144.5734,
      "step": 23990
    },
    {
      "epoch": 0.19853579848616454,
      "grad_norm": 1144.2825927734375,
      "learning_rate": 9.733587079350254e-06,
      "loss": 107.7752,
      "step": 24000
    },
    {
      "epoch": 0.19861852173553377,
      "grad_norm": 2160.167236328125,
      "learning_rate": 9.73313266057847e-06,
      "loss": 129.0158,
      "step": 24010
    },
    {
      "epoch": 0.19870124498490302,
      "grad_norm": 1064.593994140625,
      "learning_rate": 9.732677865213044e-06,
      "loss": 141.8949,
      "step": 24020
    },
    {
      "epoch": 0.19878396823427225,
      "grad_norm": 1279.8077392578125,
      "learning_rate": 9.73222269329016e-06,
      "loss": 153.0952,
      "step": 24030
    },
    {
      "epoch": 0.19886669148364147,
      "grad_norm": 950.4849853515625,
      "learning_rate": 9.731767144846034e-06,
      "loss": 146.848,
      "step": 24040
    },
    {
      "epoch": 0.1989494147330107,
      "grad_norm": 1283.0186767578125,
      "learning_rate": 9.731311219916912e-06,
      "loss": 150.6841,
      "step": 24050
    },
    {
      "epoch": 0.19903213798237995,
      "grad_norm": 704.3133544921875,
      "learning_rate": 9.730854918539072e-06,
      "loss": 149.4796,
      "step": 24060
    },
    {
      "epoch": 0.19911486123174918,
      "grad_norm": 846.5486450195312,
      "learning_rate": 9.730398240748816e-06,
      "loss": 138.2696,
      "step": 24070
    },
    {
      "epoch": 0.1991975844811184,
      "grad_norm": 779.280029296875,
      "learning_rate": 9.729941186582482e-06,
      "loss": 114.9246,
      "step": 24080
    },
    {
      "epoch": 0.19928030773048766,
      "grad_norm": 1280.6629638671875,
      "learning_rate": 9.729483756076436e-06,
      "loss": 96.8995,
      "step": 24090
    },
    {
      "epoch": 0.1993630309798569,
      "grad_norm": 626.4833374023438,
      "learning_rate": 9.729025949267072e-06,
      "loss": 117.7161,
      "step": 24100
    },
    {
      "epoch": 0.1994457542292261,
      "grad_norm": 866.7782592773438,
      "learning_rate": 9.728567766190817e-06,
      "loss": 110.1413,
      "step": 24110
    },
    {
      "epoch": 0.19952847747859537,
      "grad_norm": 1430.4652099609375,
      "learning_rate": 9.728109206884125e-06,
      "loss": 122.4254,
      "step": 24120
    },
    {
      "epoch": 0.1996112007279646,
      "grad_norm": 1178.0247802734375,
      "learning_rate": 9.727650271383485e-06,
      "loss": 131.602,
      "step": 24130
    },
    {
      "epoch": 0.19969392397733382,
      "grad_norm": 944.5925903320312,
      "learning_rate": 9.727190959725407e-06,
      "loss": 91.2385,
      "step": 24140
    },
    {
      "epoch": 0.19977664722670307,
      "grad_norm": 1111.4093017578125,
      "learning_rate": 9.72673127194644e-06,
      "loss": 185.1948,
      "step": 24150
    },
    {
      "epoch": 0.1998593704760723,
      "grad_norm": 1125.5455322265625,
      "learning_rate": 9.72627120808316e-06,
      "loss": 157.4787,
      "step": 24160
    },
    {
      "epoch": 0.19994209372544153,
      "grad_norm": 1090.2935791015625,
      "learning_rate": 9.725810768172169e-06,
      "loss": 106.9974,
      "step": 24170
    },
    {
      "epoch": 0.20002481697481078,
      "grad_norm": 620.5934448242188,
      "learning_rate": 9.725349952250105e-06,
      "loss": 115.3673,
      "step": 24180
    },
    {
      "epoch": 0.20010754022418,
      "grad_norm": 1263.1217041015625,
      "learning_rate": 9.724888760353631e-06,
      "loss": 148.2358,
      "step": 24190
    },
    {
      "epoch": 0.20019026347354923,
      "grad_norm": 1106.949951171875,
      "learning_rate": 9.72442719251944e-06,
      "loss": 170.5474,
      "step": 24200
    },
    {
      "epoch": 0.2002729867229185,
      "grad_norm": 1334.36962890625,
      "learning_rate": 9.723965248784264e-06,
      "loss": 149.0543,
      "step": 24210
    },
    {
      "epoch": 0.20035570997228772,
      "grad_norm": 1211.2431640625,
      "learning_rate": 9.723502929184851e-06,
      "loss": 126.5367,
      "step": 24220
    },
    {
      "epoch": 0.20043843322165694,
      "grad_norm": 863.3626708984375,
      "learning_rate": 9.723040233757987e-06,
      "loss": 142.1387,
      "step": 24230
    },
    {
      "epoch": 0.20052115647102617,
      "grad_norm": 1154.0657958984375,
      "learning_rate": 9.722577162540489e-06,
      "loss": 123.5952,
      "step": 24240
    },
    {
      "epoch": 0.20060387972039542,
      "grad_norm": 1012.69140625,
      "learning_rate": 9.7221137155692e-06,
      "loss": 110.8877,
      "step": 24250
    },
    {
      "epoch": 0.20068660296976465,
      "grad_norm": 1386.55859375,
      "learning_rate": 9.721649892880995e-06,
      "loss": 170.2592,
      "step": 24260
    },
    {
      "epoch": 0.20076932621913388,
      "grad_norm": 1342.9095458984375,
      "learning_rate": 9.721185694512776e-06,
      "loss": 134.4591,
      "step": 24270
    },
    {
      "epoch": 0.20085204946850313,
      "grad_norm": 392.2961120605469,
      "learning_rate": 9.720721120501478e-06,
      "loss": 134.1603,
      "step": 24280
    },
    {
      "epoch": 0.20093477271787236,
      "grad_norm": 826.9261474609375,
      "learning_rate": 9.720256170884066e-06,
      "loss": 101.2419,
      "step": 24290
    },
    {
      "epoch": 0.20101749596724158,
      "grad_norm": 791.0903930664062,
      "learning_rate": 9.719790845697534e-06,
      "loss": 129.4301,
      "step": 24300
    },
    {
      "epoch": 0.20110021921661084,
      "grad_norm": 1188.9549560546875,
      "learning_rate": 9.719325144978907e-06,
      "loss": 145.1762,
      "step": 24310
    },
    {
      "epoch": 0.20118294246598006,
      "grad_norm": 1141.0123291015625,
      "learning_rate": 9.718859068765234e-06,
      "loss": 107.9242,
      "step": 24320
    },
    {
      "epoch": 0.2012656657153493,
      "grad_norm": 1065.128173828125,
      "learning_rate": 9.718392617093602e-06,
      "loss": 113.2393,
      "step": 24330
    },
    {
      "epoch": 0.20134838896471854,
      "grad_norm": 1253.3646240234375,
      "learning_rate": 9.717925790001125e-06,
      "loss": 77.6727,
      "step": 24340
    },
    {
      "epoch": 0.20143111221408777,
      "grad_norm": 1093.85498046875,
      "learning_rate": 9.717458587524946e-06,
      "loss": 112.4173,
      "step": 24350
    },
    {
      "epoch": 0.201513835463457,
      "grad_norm": 787.4635009765625,
      "learning_rate": 9.716991009702236e-06,
      "loss": 114.5591,
      "step": 24360
    },
    {
      "epoch": 0.20159655871282625,
      "grad_norm": 926.3908081054688,
      "learning_rate": 9.7165230565702e-06,
      "loss": 123.4194,
      "step": 24370
    },
    {
      "epoch": 0.20167928196219548,
      "grad_norm": 792.05859375,
      "learning_rate": 9.71605472816607e-06,
      "loss": 128.6131,
      "step": 24380
    },
    {
      "epoch": 0.2017620052115647,
      "grad_norm": 835.9586791992188,
      "learning_rate": 9.71558602452711e-06,
      "loss": 125.1441,
      "step": 24390
    },
    {
      "epoch": 0.20184472846093396,
      "grad_norm": 1247.4454345703125,
      "learning_rate": 9.71511694569061e-06,
      "loss": 166.359,
      "step": 24400
    },
    {
      "epoch": 0.20192745171030319,
      "grad_norm": 828.7860107421875,
      "learning_rate": 9.714647491693897e-06,
      "loss": 140.4719,
      "step": 24410
    },
    {
      "epoch": 0.2020101749596724,
      "grad_norm": 843.8282470703125,
      "learning_rate": 9.714177662574316e-06,
      "loss": 101.7997,
      "step": 24420
    },
    {
      "epoch": 0.20209289820904164,
      "grad_norm": 967.5045776367188,
      "learning_rate": 9.713707458369258e-06,
      "loss": 124.0601,
      "step": 24430
    },
    {
      "epoch": 0.2021756214584109,
      "grad_norm": 692.8041381835938,
      "learning_rate": 9.713236879116127e-06,
      "loss": 120.8918,
      "step": 24440
    },
    {
      "epoch": 0.20225834470778012,
      "grad_norm": 1022.600341796875,
      "learning_rate": 9.71276592485237e-06,
      "loss": 128.2274,
      "step": 24450
    },
    {
      "epoch": 0.20234106795714935,
      "grad_norm": 1309.6890869140625,
      "learning_rate": 9.712294595615458e-06,
      "loss": 134.2852,
      "step": 24460
    },
    {
      "epoch": 0.2024237912065186,
      "grad_norm": 596.5182495117188,
      "learning_rate": 9.711822891442887e-06,
      "loss": 141.3561,
      "step": 24470
    },
    {
      "epoch": 0.20250651445588783,
      "grad_norm": 1456.3428955078125,
      "learning_rate": 9.711350812372198e-06,
      "loss": 166.3525,
      "step": 24480
    },
    {
      "epoch": 0.20258923770525705,
      "grad_norm": 1364.7294921875,
      "learning_rate": 9.710878358440945e-06,
      "loss": 178.6096,
      "step": 24490
    },
    {
      "epoch": 0.2026719609546263,
      "grad_norm": 941.8969116210938,
      "learning_rate": 9.710405529686722e-06,
      "loss": 154.1988,
      "step": 24500
    },
    {
      "epoch": 0.20275468420399553,
      "grad_norm": 1231.3597412109375,
      "learning_rate": 9.709932326147147e-06,
      "loss": 109.8311,
      "step": 24510
    },
    {
      "epoch": 0.20283740745336476,
      "grad_norm": 981.5293579101562,
      "learning_rate": 9.709458747859874e-06,
      "loss": 106.3588,
      "step": 24520
    },
    {
      "epoch": 0.20292013070273401,
      "grad_norm": 882.2429809570312,
      "learning_rate": 9.708984794862581e-06,
      "loss": 106.4644,
      "step": 24530
    },
    {
      "epoch": 0.20300285395210324,
      "grad_norm": 1536.478271484375,
      "learning_rate": 9.708510467192981e-06,
      "loss": 115.3478,
      "step": 24540
    },
    {
      "epoch": 0.20308557720147247,
      "grad_norm": 480.1463928222656,
      "learning_rate": 9.70803576488881e-06,
      "loss": 131.8311,
      "step": 24550
    },
    {
      "epoch": 0.20316830045084172,
      "grad_norm": 751.3448486328125,
      "learning_rate": 9.707560687987843e-06,
      "loss": 100.0164,
      "step": 24560
    },
    {
      "epoch": 0.20325102370021095,
      "grad_norm": 983.8947143554688,
      "learning_rate": 9.707085236527873e-06,
      "loss": 124.7845,
      "step": 24570
    },
    {
      "epoch": 0.20333374694958017,
      "grad_norm": 778.719970703125,
      "learning_rate": 9.706609410546736e-06,
      "loss": 70.2131,
      "step": 24580
    },
    {
      "epoch": 0.20341647019894943,
      "grad_norm": 799.94580078125,
      "learning_rate": 9.706133210082288e-06,
      "loss": 107.5597,
      "step": 24590
    },
    {
      "epoch": 0.20349919344831865,
      "grad_norm": 775.4839477539062,
      "learning_rate": 9.705656635172418e-06,
      "loss": 123.562,
      "step": 24600
    },
    {
      "epoch": 0.20358191669768788,
      "grad_norm": 906.2435302734375,
      "learning_rate": 9.705179685855048e-06,
      "loss": 141.3586,
      "step": 24610
    },
    {
      "epoch": 0.2036646399470571,
      "grad_norm": 572.1375732421875,
      "learning_rate": 9.704702362168121e-06,
      "loss": 174.9464,
      "step": 24620
    },
    {
      "epoch": 0.20374736319642636,
      "grad_norm": 1068.9876708984375,
      "learning_rate": 9.704224664149621e-06,
      "loss": 122.0626,
      "step": 24630
    },
    {
      "epoch": 0.2038300864457956,
      "grad_norm": 606.3220825195312,
      "learning_rate": 9.703746591837552e-06,
      "loss": 109.6907,
      "step": 24640
    },
    {
      "epoch": 0.20391280969516482,
      "grad_norm": 1032.2161865234375,
      "learning_rate": 9.703268145269957e-06,
      "loss": 126.0382,
      "step": 24650
    },
    {
      "epoch": 0.20399553294453407,
      "grad_norm": 1022.0555419921875,
      "learning_rate": 9.702789324484898e-06,
      "loss": 188.598,
      "step": 24660
    },
    {
      "epoch": 0.2040782561939033,
      "grad_norm": 1047.5244140625,
      "learning_rate": 9.702310129520476e-06,
      "loss": 122.4435,
      "step": 24670
    },
    {
      "epoch": 0.20416097944327252,
      "grad_norm": 1051.2161865234375,
      "learning_rate": 9.701830560414817e-06,
      "loss": 144.9207,
      "step": 24680
    },
    {
      "epoch": 0.20424370269264178,
      "grad_norm": 1291.77099609375,
      "learning_rate": 9.701350617206081e-06,
      "loss": 141.4524,
      "step": 24690
    },
    {
      "epoch": 0.204326425942011,
      "grad_norm": 1048.1988525390625,
      "learning_rate": 9.700870299932453e-06,
      "loss": 133.1601,
      "step": 24700
    },
    {
      "epoch": 0.20440914919138023,
      "grad_norm": 1616.698486328125,
      "learning_rate": 9.700389608632146e-06,
      "loss": 159.5847,
      "step": 24710
    },
    {
      "epoch": 0.20449187244074948,
      "grad_norm": 863.6107788085938,
      "learning_rate": 9.699908543343413e-06,
      "loss": 135.6566,
      "step": 24720
    },
    {
      "epoch": 0.2045745956901187,
      "grad_norm": 1772.513916015625,
      "learning_rate": 9.699427104104525e-06,
      "loss": 202.6929,
      "step": 24730
    },
    {
      "epoch": 0.20465731893948794,
      "grad_norm": 822.7219848632812,
      "learning_rate": 9.698945290953789e-06,
      "loss": 140.5493,
      "step": 24740
    },
    {
      "epoch": 0.2047400421888572,
      "grad_norm": 811.2693481445312,
      "learning_rate": 9.698463103929542e-06,
      "loss": 81.8386,
      "step": 24750
    },
    {
      "epoch": 0.20482276543822642,
      "grad_norm": 530.8605346679688,
      "learning_rate": 9.69798054307015e-06,
      "loss": 106.7644,
      "step": 24760
    },
    {
      "epoch": 0.20490548868759564,
      "grad_norm": 2197.434326171875,
      "learning_rate": 9.697497608414007e-06,
      "loss": 189.2195,
      "step": 24770
    },
    {
      "epoch": 0.20498821193696487,
      "grad_norm": 1628.6551513671875,
      "learning_rate": 9.697014299999536e-06,
      "loss": 110.0945,
      "step": 24780
    },
    {
      "epoch": 0.20507093518633412,
      "grad_norm": 988.3732299804688,
      "learning_rate": 9.696530617865197e-06,
      "loss": 147.2918,
      "step": 24790
    },
    {
      "epoch": 0.20515365843570335,
      "grad_norm": 934.0083618164062,
      "learning_rate": 9.696046562049469e-06,
      "loss": 124.3846,
      "step": 24800
    },
    {
      "epoch": 0.20523638168507258,
      "grad_norm": 1064.4521484375,
      "learning_rate": 9.695562132590865e-06,
      "loss": 130.5005,
      "step": 24810
    },
    {
      "epoch": 0.20531910493444183,
      "grad_norm": 936.8753662109375,
      "learning_rate": 9.695077329527936e-06,
      "loss": 79.4409,
      "step": 24820
    },
    {
      "epoch": 0.20540182818381106,
      "grad_norm": 2023.431396484375,
      "learning_rate": 9.694592152899249e-06,
      "loss": 135.3671,
      "step": 24830
    },
    {
      "epoch": 0.20548455143318028,
      "grad_norm": 1526.9307861328125,
      "learning_rate": 9.694106602743411e-06,
      "loss": 141.677,
      "step": 24840
    },
    {
      "epoch": 0.20556727468254954,
      "grad_norm": 1296.6025390625,
      "learning_rate": 9.693620679099055e-06,
      "loss": 106.9513,
      "step": 24850
    },
    {
      "epoch": 0.20564999793191877,
      "grad_norm": 675.1536865234375,
      "learning_rate": 9.693134382004839e-06,
      "loss": 122.0934,
      "step": 24860
    },
    {
      "epoch": 0.205732721181288,
      "grad_norm": 1233.307373046875,
      "learning_rate": 9.69264771149946e-06,
      "loss": 152.0633,
      "step": 24870
    },
    {
      "epoch": 0.20581544443065725,
      "grad_norm": 2084.801025390625,
      "learning_rate": 9.692160667621639e-06,
      "loss": 191.7889,
      "step": 24880
    },
    {
      "epoch": 0.20589816768002647,
      "grad_norm": 1063.7586669921875,
      "learning_rate": 9.69167325041013e-06,
      "loss": 98.7458,
      "step": 24890
    },
    {
      "epoch": 0.2059808909293957,
      "grad_norm": 1160.4176025390625,
      "learning_rate": 9.69118545990371e-06,
      "loss": 136.4698,
      "step": 24900
    },
    {
      "epoch": 0.20606361417876495,
      "grad_norm": 1240.2447509765625,
      "learning_rate": 9.690697296141194e-06,
      "loss": 131.5934,
      "step": 24910
    },
    {
      "epoch": 0.20614633742813418,
      "grad_norm": 1285.3697509765625,
      "learning_rate": 9.690208759161418e-06,
      "loss": 143.4328,
      "step": 24920
    },
    {
      "epoch": 0.2062290606775034,
      "grad_norm": 899.1880493164062,
      "learning_rate": 9.689719849003261e-06,
      "loss": 150.501,
      "step": 24930
    },
    {
      "epoch": 0.20631178392687266,
      "grad_norm": 1172.1856689453125,
      "learning_rate": 9.689230565705617e-06,
      "loss": 88.1189,
      "step": 24940
    },
    {
      "epoch": 0.2063945071762419,
      "grad_norm": 940.1471557617188,
      "learning_rate": 9.688740909307416e-06,
      "loss": 126.9263,
      "step": 24950
    },
    {
      "epoch": 0.2064772304256111,
      "grad_norm": 1031.7501220703125,
      "learning_rate": 9.68825087984762e-06,
      "loss": 145.9387,
      "step": 24960
    },
    {
      "epoch": 0.20655995367498034,
      "grad_norm": 747.2893676757812,
      "learning_rate": 9.687760477365217e-06,
      "loss": 111.3116,
      "step": 24970
    },
    {
      "epoch": 0.2066426769243496,
      "grad_norm": 1009.5612182617188,
      "learning_rate": 9.687269701899228e-06,
      "loss": 122.0988,
      "step": 24980
    },
    {
      "epoch": 0.20672540017371882,
      "grad_norm": 1089.2601318359375,
      "learning_rate": 9.6867785534887e-06,
      "loss": 118.2881,
      "step": 24990
    },
    {
      "epoch": 0.20680812342308805,
      "grad_norm": 1918.05908203125,
      "learning_rate": 9.686287032172712e-06,
      "loss": 119.0276,
      "step": 25000
    },
    {
      "epoch": 0.2068908466724573,
      "grad_norm": 1754.23486328125,
      "learning_rate": 9.685795137990372e-06,
      "loss": 149.8688,
      "step": 25010
    },
    {
      "epoch": 0.20697356992182653,
      "grad_norm": 701.3511352539062,
      "learning_rate": 9.685302870980819e-06,
      "loss": 111.5625,
      "step": 25020
    },
    {
      "epoch": 0.20705629317119575,
      "grad_norm": 1214.149658203125,
      "learning_rate": 9.684810231183218e-06,
      "loss": 104.316,
      "step": 25030
    },
    {
      "epoch": 0.207139016420565,
      "grad_norm": 1371.7374267578125,
      "learning_rate": 9.684317218636767e-06,
      "loss": 137.7989,
      "step": 25040
    },
    {
      "epoch": 0.20722173966993424,
      "grad_norm": 976.9849243164062,
      "learning_rate": 9.683823833380692e-06,
      "loss": 107.9419,
      "step": 25050
    },
    {
      "epoch": 0.20730446291930346,
      "grad_norm": 2080.160400390625,
      "learning_rate": 9.683330075454252e-06,
      "loss": 134.3795,
      "step": 25060
    },
    {
      "epoch": 0.20738718616867272,
      "grad_norm": 1066.323974609375,
      "learning_rate": 9.68283594489673e-06,
      "loss": 113.7562,
      "step": 25070
    },
    {
      "epoch": 0.20746990941804194,
      "grad_norm": 1623.5059814453125,
      "learning_rate": 9.682341441747446e-06,
      "loss": 124.4294,
      "step": 25080
    },
    {
      "epoch": 0.20755263266741117,
      "grad_norm": 1231.952880859375,
      "learning_rate": 9.68184656604574e-06,
      "loss": 168.1638,
      "step": 25090
    },
    {
      "epoch": 0.20763535591678042,
      "grad_norm": 604.9308471679688,
      "learning_rate": 9.681351317830991e-06,
      "loss": 150.3729,
      "step": 25100
    },
    {
      "epoch": 0.20771807916614965,
      "grad_norm": 915.5381469726562,
      "learning_rate": 9.680855697142601e-06,
      "loss": 89.7146,
      "step": 25110
    },
    {
      "epoch": 0.20780080241551888,
      "grad_norm": 517.771484375,
      "learning_rate": 9.680359704020005e-06,
      "loss": 110.3232,
      "step": 25120
    },
    {
      "epoch": 0.20788352566488813,
      "grad_norm": 999.2185668945312,
      "learning_rate": 9.67986333850267e-06,
      "loss": 155.6853,
      "step": 25130
    },
    {
      "epoch": 0.20796624891425736,
      "grad_norm": 837.3745727539062,
      "learning_rate": 9.679366600630085e-06,
      "loss": 123.4482,
      "step": 25140
    },
    {
      "epoch": 0.20804897216362658,
      "grad_norm": 1082.25146484375,
      "learning_rate": 9.678869490441775e-06,
      "loss": 92.413,
      "step": 25150
    },
    {
      "epoch": 0.2081316954129958,
      "grad_norm": 1187.2119140625,
      "learning_rate": 9.678372007977292e-06,
      "loss": 130.2228,
      "step": 25160
    },
    {
      "epoch": 0.20821441866236506,
      "grad_norm": 1067.64306640625,
      "learning_rate": 9.67787415327622e-06,
      "loss": 106.1636,
      "step": 25170
    },
    {
      "epoch": 0.2082971419117343,
      "grad_norm": 1119.31640625,
      "learning_rate": 9.67737592637817e-06,
      "loss": 125.416,
      "step": 25180
    },
    {
      "epoch": 0.20837986516110352,
      "grad_norm": 1132.6695556640625,
      "learning_rate": 9.676877327322785e-06,
      "loss": 121.1855,
      "step": 25190
    },
    {
      "epoch": 0.20846258841047277,
      "grad_norm": 921.9434204101562,
      "learning_rate": 9.676378356149733e-06,
      "loss": 119.0814,
      "step": 25200
    },
    {
      "epoch": 0.208545311659842,
      "grad_norm": 1111.28857421875,
      "learning_rate": 9.675879012898719e-06,
      "loss": 112.7059,
      "step": 25210
    },
    {
      "epoch": 0.20862803490921122,
      "grad_norm": 952.7010498046875,
      "learning_rate": 9.67537929760947e-06,
      "loss": 135.5189,
      "step": 25220
    },
    {
      "epoch": 0.20871075815858048,
      "grad_norm": 668.4132080078125,
      "learning_rate": 9.674879210321747e-06,
      "loss": 126.6339,
      "step": 25230
    },
    {
      "epoch": 0.2087934814079497,
      "grad_norm": 1562.3443603515625,
      "learning_rate": 9.67437875107534e-06,
      "loss": 157.3701,
      "step": 25240
    },
    {
      "epoch": 0.20887620465731893,
      "grad_norm": 787.549072265625,
      "learning_rate": 9.673877919910069e-06,
      "loss": 144.3811,
      "step": 25250
    },
    {
      "epoch": 0.20895892790668819,
      "grad_norm": 1168.210693359375,
      "learning_rate": 9.673376716865781e-06,
      "loss": 147.3342,
      "step": 25260
    },
    {
      "epoch": 0.2090416511560574,
      "grad_norm": 922.6990356445312,
      "learning_rate": 9.672875141982358e-06,
      "loss": 134.1699,
      "step": 25270
    },
    {
      "epoch": 0.20912437440542664,
      "grad_norm": 920.8440551757812,
      "learning_rate": 9.672373195299704e-06,
      "loss": 90.3388,
      "step": 25280
    },
    {
      "epoch": 0.2092070976547959,
      "grad_norm": 742.2246704101562,
      "learning_rate": 9.67187087685776e-06,
      "loss": 127.1734,
      "step": 25290
    },
    {
      "epoch": 0.20928982090416512,
      "grad_norm": 1413.235107421875,
      "learning_rate": 9.671368186696488e-06,
      "loss": 121.8572,
      "step": 25300
    },
    {
      "epoch": 0.20937254415353435,
      "grad_norm": 1089.089111328125,
      "learning_rate": 9.670865124855889e-06,
      "loss": 155.726,
      "step": 25310
    },
    {
      "epoch": 0.2094552674029036,
      "grad_norm": 1251.3277587890625,
      "learning_rate": 9.67036169137599e-06,
      "loss": 134.8554,
      "step": 25320
    },
    {
      "epoch": 0.20953799065227283,
      "grad_norm": 975.4923706054688,
      "learning_rate": 9.669857886296842e-06,
      "loss": 137.952,
      "step": 25330
    },
    {
      "epoch": 0.20962071390164205,
      "grad_norm": 1131.67822265625,
      "learning_rate": 9.669353709658537e-06,
      "loss": 119.2049,
      "step": 25340
    },
    {
      "epoch": 0.20970343715101128,
      "grad_norm": 637.6780395507812,
      "learning_rate": 9.668849161501186e-06,
      "loss": 104.5548,
      "step": 25350
    },
    {
      "epoch": 0.20978616040038053,
      "grad_norm": 1121.5142822265625,
      "learning_rate": 9.668344241864934e-06,
      "loss": 115.9487,
      "step": 25360
    },
    {
      "epoch": 0.20986888364974976,
      "grad_norm": 937.0314331054688,
      "learning_rate": 9.667838950789957e-06,
      "loss": 112.8814,
      "step": 25370
    },
    {
      "epoch": 0.209951606899119,
      "grad_norm": 1264.4849853515625,
      "learning_rate": 9.667333288316454e-06,
      "loss": 135.4153,
      "step": 25380
    },
    {
      "epoch": 0.21003433014848824,
      "grad_norm": 495.7436828613281,
      "learning_rate": 9.666827254484663e-06,
      "loss": 111.5311,
      "step": 25390
    },
    {
      "epoch": 0.21011705339785747,
      "grad_norm": 853.3590087890625,
      "learning_rate": 9.666320849334846e-06,
      "loss": 102.2723,
      "step": 25400
    },
    {
      "epoch": 0.2101997766472267,
      "grad_norm": 1231.4791259765625,
      "learning_rate": 9.665814072907293e-06,
      "loss": 118.1443,
      "step": 25410
    },
    {
      "epoch": 0.21028249989659595,
      "grad_norm": 688.8372192382812,
      "learning_rate": 9.665306925242329e-06,
      "loss": 120.3173,
      "step": 25420
    },
    {
      "epoch": 0.21036522314596517,
      "grad_norm": 1288.0115966796875,
      "learning_rate": 9.664799406380302e-06,
      "loss": 129.2124,
      "step": 25430
    },
    {
      "epoch": 0.2104479463953344,
      "grad_norm": 734.3385009765625,
      "learning_rate": 9.664291516361597e-06,
      "loss": 141.6067,
      "step": 25440
    },
    {
      "epoch": 0.21053066964470366,
      "grad_norm": 808.9627685546875,
      "learning_rate": 9.663783255226622e-06,
      "loss": 128.5043,
      "step": 25450
    },
    {
      "epoch": 0.21061339289407288,
      "grad_norm": 2131.609619140625,
      "learning_rate": 9.663274623015816e-06,
      "loss": 130.3591,
      "step": 25460
    },
    {
      "epoch": 0.2106961161434421,
      "grad_norm": 1364.452880859375,
      "learning_rate": 9.662765619769651e-06,
      "loss": 186.9397,
      "step": 25470
    },
    {
      "epoch": 0.21077883939281136,
      "grad_norm": 1547.8067626953125,
      "learning_rate": 9.662256245528622e-06,
      "loss": 130.2646,
      "step": 25480
    },
    {
      "epoch": 0.2108615626421806,
      "grad_norm": 760.1724853515625,
      "learning_rate": 9.661746500333265e-06,
      "loss": 107.9425,
      "step": 25490
    },
    {
      "epoch": 0.21094428589154982,
      "grad_norm": 691.5214233398438,
      "learning_rate": 9.66123638422413e-06,
      "loss": 131.0554,
      "step": 25500
    },
    {
      "epoch": 0.21102700914091904,
      "grad_norm": 1053.4617919921875,
      "learning_rate": 9.66072589724181e-06,
      "loss": 139.2496,
      "step": 25510
    },
    {
      "epoch": 0.2111097323902883,
      "grad_norm": 903.3119506835938,
      "learning_rate": 9.66021503942692e-06,
      "loss": 103.5156,
      "step": 25520
    },
    {
      "epoch": 0.21119245563965752,
      "grad_norm": 1562.44482421875,
      "learning_rate": 9.659703810820105e-06,
      "loss": 154.1499,
      "step": 25530
    },
    {
      "epoch": 0.21127517888902675,
      "grad_norm": 724.2481079101562,
      "learning_rate": 9.659192211462043e-06,
      "loss": 122.7316,
      "step": 25540
    },
    {
      "epoch": 0.211357902138396,
      "grad_norm": 1171.6414794921875,
      "learning_rate": 9.658680241393441e-06,
      "loss": 112.3694,
      "step": 25550
    },
    {
      "epoch": 0.21144062538776523,
      "grad_norm": 687.9328002929688,
      "learning_rate": 9.658167900655032e-06,
      "loss": 85.1699,
      "step": 25560
    },
    {
      "epoch": 0.21152334863713446,
      "grad_norm": 1203.814208984375,
      "learning_rate": 9.657655189287582e-06,
      "loss": 111.8283,
      "step": 25570
    },
    {
      "epoch": 0.2116060718865037,
      "grad_norm": 2333.49365234375,
      "learning_rate": 9.657142107331883e-06,
      "loss": 168.9521,
      "step": 25580
    },
    {
      "epoch": 0.21168879513587294,
      "grad_norm": 919.4609375,
      "learning_rate": 9.65662865482876e-06,
      "loss": 110.3846,
      "step": 25590
    },
    {
      "epoch": 0.21177151838524216,
      "grad_norm": 1061.5003662109375,
      "learning_rate": 9.656114831819067e-06,
      "loss": 133.6754,
      "step": 25600
    },
    {
      "epoch": 0.21185424163461142,
      "grad_norm": 498.1741638183594,
      "learning_rate": 9.655600638343685e-06,
      "loss": 109.6762,
      "step": 25610
    },
    {
      "epoch": 0.21193696488398064,
      "grad_norm": 946.8280639648438,
      "learning_rate": 9.655086074443527e-06,
      "loss": 183.0106,
      "step": 25620
    },
    {
      "epoch": 0.21201968813334987,
      "grad_norm": 1837.9036865234375,
      "learning_rate": 9.654571140159534e-06,
      "loss": 122.9791,
      "step": 25630
    },
    {
      "epoch": 0.21210241138271912,
      "grad_norm": 1461.47900390625,
      "learning_rate": 9.654055835532676e-06,
      "loss": 191.4103,
      "step": 25640
    },
    {
      "epoch": 0.21218513463208835,
      "grad_norm": 820.8297119140625,
      "learning_rate": 9.653540160603956e-06,
      "loss": 115.1019,
      "step": 25650
    },
    {
      "epoch": 0.21226785788145758,
      "grad_norm": 1157.8956298828125,
      "learning_rate": 9.653024115414402e-06,
      "loss": 142.3736,
      "step": 25660
    },
    {
      "epoch": 0.21235058113082683,
      "grad_norm": 917.7392578125,
      "learning_rate": 9.652507700005072e-06,
      "loss": 130.4384,
      "step": 25670
    },
    {
      "epoch": 0.21243330438019606,
      "grad_norm": 1544.7633056640625,
      "learning_rate": 9.651990914417057e-06,
      "loss": 135.7345,
      "step": 25680
    },
    {
      "epoch": 0.21251602762956529,
      "grad_norm": 768.8818359375,
      "learning_rate": 9.651473758691477e-06,
      "loss": 99.1156,
      "step": 25690
    },
    {
      "epoch": 0.2125987508789345,
      "grad_norm": 1055.4837646484375,
      "learning_rate": 9.650956232869475e-06,
      "loss": 113.1447,
      "step": 25700
    },
    {
      "epoch": 0.21268147412830377,
      "grad_norm": 1085.100341796875,
      "learning_rate": 9.650438336992231e-06,
      "loss": 156.4204,
      "step": 25710
    },
    {
      "epoch": 0.212764197377673,
      "grad_norm": 535.8530883789062,
      "learning_rate": 9.64992007110095e-06,
      "loss": 113.1042,
      "step": 25720
    },
    {
      "epoch": 0.21284692062704222,
      "grad_norm": 1140.1822509765625,
      "learning_rate": 9.64940143523687e-06,
      "loss": 134.6167,
      "step": 25730
    },
    {
      "epoch": 0.21292964387641147,
      "grad_norm": 1358.8885498046875,
      "learning_rate": 9.648882429441258e-06,
      "loss": 133.8381,
      "step": 25740
    },
    {
      "epoch": 0.2130123671257807,
      "grad_norm": 1481.0440673828125,
      "learning_rate": 9.648363053755406e-06,
      "loss": 120.2028,
      "step": 25750
    },
    {
      "epoch": 0.21309509037514993,
      "grad_norm": 1080.2623291015625,
      "learning_rate": 9.647843308220636e-06,
      "loss": 105.3537,
      "step": 25760
    },
    {
      "epoch": 0.21317781362451918,
      "grad_norm": 712.189453125,
      "learning_rate": 9.647323192878306e-06,
      "loss": 101.1071,
      "step": 25770
    },
    {
      "epoch": 0.2132605368738884,
      "grad_norm": 1282.9239501953125,
      "learning_rate": 9.646802707769798e-06,
      "loss": 121.2276,
      "step": 25780
    },
    {
      "epoch": 0.21334326012325763,
      "grad_norm": 558.678955078125,
      "learning_rate": 9.646281852936525e-06,
      "loss": 92.6775,
      "step": 25790
    },
    {
      "epoch": 0.2134259833726269,
      "grad_norm": 994.920654296875,
      "learning_rate": 9.64576062841993e-06,
      "loss": 124.3649,
      "step": 25800
    },
    {
      "epoch": 0.2135087066219961,
      "grad_norm": 1420.953857421875,
      "learning_rate": 9.64523903426148e-06,
      "loss": 122.571,
      "step": 25810
    },
    {
      "epoch": 0.21359142987136534,
      "grad_norm": 859.2244873046875,
      "learning_rate": 9.64471707050268e-06,
      "loss": 126.6968,
      "step": 25820
    },
    {
      "epoch": 0.2136741531207346,
      "grad_norm": 1234.9310302734375,
      "learning_rate": 9.644194737185058e-06,
      "loss": 139.6012,
      "step": 25830
    },
    {
      "epoch": 0.21375687637010382,
      "grad_norm": 1804.944091796875,
      "learning_rate": 9.643672034350177e-06,
      "loss": 150.6442,
      "step": 25840
    },
    {
      "epoch": 0.21383959961947305,
      "grad_norm": 1387.7882080078125,
      "learning_rate": 9.643148962039622e-06,
      "loss": 93.8409,
      "step": 25850
    },
    {
      "epoch": 0.2139223228688423,
      "grad_norm": 1455.9395751953125,
      "learning_rate": 9.642625520295014e-06,
      "loss": 174.9808,
      "step": 25860
    },
    {
      "epoch": 0.21400504611821153,
      "grad_norm": 1184.424072265625,
      "learning_rate": 9.642101709158001e-06,
      "loss": 129.4004,
      "step": 25870
    },
    {
      "epoch": 0.21408776936758075,
      "grad_norm": 1059.5274658203125,
      "learning_rate": 9.641577528670257e-06,
      "loss": 129.7015,
      "step": 25880
    },
    {
      "epoch": 0.21417049261694998,
      "grad_norm": 753.4532470703125,
      "learning_rate": 9.641052978873494e-06,
      "loss": 109.8265,
      "step": 25890
    },
    {
      "epoch": 0.21425321586631924,
      "grad_norm": 676.7515258789062,
      "learning_rate": 9.640528059809442e-06,
      "loss": 74.3417,
      "step": 25900
    },
    {
      "epoch": 0.21433593911568846,
      "grad_norm": 2531.762939453125,
      "learning_rate": 9.640002771519872e-06,
      "loss": 133.8275,
      "step": 25910
    },
    {
      "epoch": 0.2144186623650577,
      "grad_norm": 1403.41943359375,
      "learning_rate": 9.639477114046575e-06,
      "loss": 164.8685,
      "step": 25920
    },
    {
      "epoch": 0.21450138561442694,
      "grad_norm": 922.8995971679688,
      "learning_rate": 9.638951087431376e-06,
      "loss": 114.9469,
      "step": 25930
    },
    {
      "epoch": 0.21458410886379617,
      "grad_norm": 1005.201416015625,
      "learning_rate": 9.638424691716129e-06,
      "loss": 144.4065,
      "step": 25940
    },
    {
      "epoch": 0.2146668321131654,
      "grad_norm": 917.942626953125,
      "learning_rate": 9.637897926942716e-06,
      "loss": 140.8008,
      "step": 25950
    },
    {
      "epoch": 0.21474955536253465,
      "grad_norm": 1009.6735229492188,
      "learning_rate": 9.637370793153051e-06,
      "loss": 132.449,
      "step": 25960
    },
    {
      "epoch": 0.21483227861190388,
      "grad_norm": 904.17529296875,
      "learning_rate": 9.636843290389076e-06,
      "loss": 114.3653,
      "step": 25970
    },
    {
      "epoch": 0.2149150018612731,
      "grad_norm": 1091.709716796875,
      "learning_rate": 9.636315418692759e-06,
      "loss": 108.0948,
      "step": 25980
    },
    {
      "epoch": 0.21499772511064236,
      "grad_norm": 2188.771240234375,
      "learning_rate": 9.635787178106102e-06,
      "loss": 107.5951,
      "step": 25990
    },
    {
      "epoch": 0.21508044836001158,
      "grad_norm": 1075.4591064453125,
      "learning_rate": 9.635258568671135e-06,
      "loss": 137.4553,
      "step": 26000
    },
    {
      "epoch": 0.2151631716093808,
      "grad_norm": 916.110107421875,
      "learning_rate": 9.634729590429917e-06,
      "loss": 123.9193,
      "step": 26010
    },
    {
      "epoch": 0.21524589485875006,
      "grad_norm": 1345.30615234375,
      "learning_rate": 9.634200243424535e-06,
      "loss": 139.9196,
      "step": 26020
    },
    {
      "epoch": 0.2153286181081193,
      "grad_norm": 1103.3697509765625,
      "learning_rate": 9.633670527697108e-06,
      "loss": 130.9413,
      "step": 26030
    },
    {
      "epoch": 0.21541134135748852,
      "grad_norm": 1355.6485595703125,
      "learning_rate": 9.633140443289784e-06,
      "loss": 190.8187,
      "step": 26040
    },
    {
      "epoch": 0.21549406460685774,
      "grad_norm": 575.96142578125,
      "learning_rate": 9.632609990244737e-06,
      "loss": 103.7051,
      "step": 26050
    },
    {
      "epoch": 0.215576787856227,
      "grad_norm": 779.9686889648438,
      "learning_rate": 9.632079168604175e-06,
      "loss": 109.2936,
      "step": 26060
    },
    {
      "epoch": 0.21565951110559622,
      "grad_norm": 1010.6686401367188,
      "learning_rate": 9.63154797841033e-06,
      "loss": 159.4392,
      "step": 26070
    },
    {
      "epoch": 0.21574223435496545,
      "grad_norm": 826.314208984375,
      "learning_rate": 9.63101641970547e-06,
      "loss": 129.3246,
      "step": 26080
    },
    {
      "epoch": 0.2158249576043347,
      "grad_norm": 898.4963989257812,
      "learning_rate": 9.630484492531886e-06,
      "loss": 129.947,
      "step": 26090
    },
    {
      "epoch": 0.21590768085370393,
      "grad_norm": 843.03857421875,
      "learning_rate": 9.629952196931902e-06,
      "loss": 113.0155,
      "step": 26100
    },
    {
      "epoch": 0.21599040410307316,
      "grad_norm": 660.1622314453125,
      "learning_rate": 9.629419532947872e-06,
      "loss": 156.4598,
      "step": 26110
    },
    {
      "epoch": 0.2160731273524424,
      "grad_norm": 1190.5133056640625,
      "learning_rate": 9.628886500622174e-06,
      "loss": 128.8638,
      "step": 26120
    },
    {
      "epoch": 0.21615585060181164,
      "grad_norm": 924.6058959960938,
      "learning_rate": 9.62835309999722e-06,
      "loss": 99.8876,
      "step": 26130
    },
    {
      "epoch": 0.21623857385118087,
      "grad_norm": 1131.3111572265625,
      "learning_rate": 9.627819331115453e-06,
      "loss": 126.6344,
      "step": 26140
    },
    {
      "epoch": 0.21632129710055012,
      "grad_norm": 784.7423706054688,
      "learning_rate": 9.627285194019342e-06,
      "loss": 102.5163,
      "step": 26150
    },
    {
      "epoch": 0.21640402034991935,
      "grad_norm": 1271.62890625,
      "learning_rate": 9.626750688751382e-06,
      "loss": 115.6172,
      "step": 26160
    },
    {
      "epoch": 0.21648674359928857,
      "grad_norm": 1133.4263916015625,
      "learning_rate": 9.626215815354104e-06,
      "loss": 98.0378,
      "step": 26170
    },
    {
      "epoch": 0.21656946684865783,
      "grad_norm": 822.8490600585938,
      "learning_rate": 9.625680573870067e-06,
      "loss": 123.6515,
      "step": 26180
    },
    {
      "epoch": 0.21665219009802705,
      "grad_norm": 796.753662109375,
      "learning_rate": 9.625144964341853e-06,
      "loss": 127.32,
      "step": 26190
    },
    {
      "epoch": 0.21673491334739628,
      "grad_norm": 976.165771484375,
      "learning_rate": 9.624608986812082e-06,
      "loss": 113.2206,
      "step": 26200
    },
    {
      "epoch": 0.21681763659676553,
      "grad_norm": 1549.4820556640625,
      "learning_rate": 9.624072641323398e-06,
      "loss": 121.0571,
      "step": 26210
    },
    {
      "epoch": 0.21690035984613476,
      "grad_norm": 1164.861328125,
      "learning_rate": 9.623535927918474e-06,
      "loss": 168.508,
      "step": 26220
    },
    {
      "epoch": 0.216983083095504,
      "grad_norm": 1254.7081298828125,
      "learning_rate": 9.622998846640018e-06,
      "loss": 114.7848,
      "step": 26230
    },
    {
      "epoch": 0.2170658063448732,
      "grad_norm": 744.2569580078125,
      "learning_rate": 9.62246139753076e-06,
      "loss": 125.7201,
      "step": 26240
    },
    {
      "epoch": 0.21714852959424247,
      "grad_norm": 1193.7801513671875,
      "learning_rate": 9.621923580633462e-06,
      "loss": 109.2309,
      "step": 26250
    },
    {
      "epoch": 0.2172312528436117,
      "grad_norm": 933.6427612304688,
      "learning_rate": 9.621385395990915e-06,
      "loss": 137.9964,
      "step": 26260
    },
    {
      "epoch": 0.21731397609298092,
      "grad_norm": 1021.1292724609375,
      "learning_rate": 9.620846843645944e-06,
      "loss": 105.2249,
      "step": 26270
    },
    {
      "epoch": 0.21739669934235017,
      "grad_norm": 1435.4403076171875,
      "learning_rate": 9.620307923641395e-06,
      "loss": 120.3464,
      "step": 26280
    },
    {
      "epoch": 0.2174794225917194,
      "grad_norm": 1647.5338134765625,
      "learning_rate": 9.61976863602015e-06,
      "loss": 115.3988,
      "step": 26290
    },
    {
      "epoch": 0.21756214584108863,
      "grad_norm": 729.9489135742188,
      "learning_rate": 9.619228980825114e-06,
      "loss": 171.7545,
      "step": 26300
    },
    {
      "epoch": 0.21764486909045788,
      "grad_norm": 1271.6209716796875,
      "learning_rate": 9.61868895809923e-06,
      "loss": 120.3744,
      "step": 26310
    },
    {
      "epoch": 0.2177275923398271,
      "grad_norm": 844.6718139648438,
      "learning_rate": 9.618148567885462e-06,
      "loss": 142.1199,
      "step": 26320
    },
    {
      "epoch": 0.21781031558919633,
      "grad_norm": 837.2586669921875,
      "learning_rate": 9.617607810226806e-06,
      "loss": 154.3331,
      "step": 26330
    },
    {
      "epoch": 0.2178930388385656,
      "grad_norm": 1400.4326171875,
      "learning_rate": 9.61706668516629e-06,
      "loss": 124.3974,
      "step": 26340
    },
    {
      "epoch": 0.21797576208793482,
      "grad_norm": 1115.8116455078125,
      "learning_rate": 9.616525192746965e-06,
      "loss": 105.8458,
      "step": 26350
    },
    {
      "epoch": 0.21805848533730404,
      "grad_norm": 1421.916748046875,
      "learning_rate": 9.61598333301192e-06,
      "loss": 156.9738,
      "step": 26360
    },
    {
      "epoch": 0.2181412085866733,
      "grad_norm": 798.7007446289062,
      "learning_rate": 9.615441106004264e-06,
      "loss": 111.3176,
      "step": 26370
    },
    {
      "epoch": 0.21822393183604252,
      "grad_norm": 1311.8187255859375,
      "learning_rate": 9.614898511767142e-06,
      "loss": 112.8957,
      "step": 26380
    },
    {
      "epoch": 0.21830665508541175,
      "grad_norm": 1265.518798828125,
      "learning_rate": 9.614355550343724e-06,
      "loss": 97.6749,
      "step": 26390
    },
    {
      "epoch": 0.218389378334781,
      "grad_norm": 1181.8995361328125,
      "learning_rate": 9.613812221777212e-06,
      "loss": 126.788,
      "step": 26400
    },
    {
      "epoch": 0.21847210158415023,
      "grad_norm": 921.5441284179688,
      "learning_rate": 9.613268526110838e-06,
      "loss": 126.4273,
      "step": 26410
    },
    {
      "epoch": 0.21855482483351946,
      "grad_norm": 2307.983642578125,
      "learning_rate": 9.612724463387857e-06,
      "loss": 124.0576,
      "step": 26420
    },
    {
      "epoch": 0.21863754808288868,
      "grad_norm": 1391.1617431640625,
      "learning_rate": 9.612180033651561e-06,
      "loss": 122.5418,
      "step": 26430
    },
    {
      "epoch": 0.21872027133225794,
      "grad_norm": 883.1130981445312,
      "learning_rate": 9.611635236945267e-06,
      "loss": 105.9835,
      "step": 26440
    },
    {
      "epoch": 0.21880299458162716,
      "grad_norm": 1083.864990234375,
      "learning_rate": 9.61109007331232e-06,
      "loss": 124.8244,
      "step": 26450
    },
    {
      "epoch": 0.2188857178309964,
      "grad_norm": 1159.39697265625,
      "learning_rate": 9.610544542796101e-06,
      "loss": 120.3728,
      "step": 26460
    },
    {
      "epoch": 0.21896844108036564,
      "grad_norm": 1433.8590087890625,
      "learning_rate": 9.609998645440011e-06,
      "loss": 132.6535,
      "step": 26470
    },
    {
      "epoch": 0.21905116432973487,
      "grad_norm": 1183.7196044921875,
      "learning_rate": 9.609452381287486e-06,
      "loss": 133.8586,
      "step": 26480
    },
    {
      "epoch": 0.2191338875791041,
      "grad_norm": 859.5003662109375,
      "learning_rate": 9.608905750381988e-06,
      "loss": 101.3727,
      "step": 26490
    },
    {
      "epoch": 0.21921661082847335,
      "grad_norm": 800.9340209960938,
      "learning_rate": 9.608358752767013e-06,
      "loss": 142.6461,
      "step": 26500
    },
    {
      "epoch": 0.21929933407784258,
      "grad_norm": 1471.0047607421875,
      "learning_rate": 9.60781138848608e-06,
      "loss": 110.0766,
      "step": 26510
    },
    {
      "epoch": 0.2193820573272118,
      "grad_norm": 954.5092163085938,
      "learning_rate": 9.607263657582744e-06,
      "loss": 110.3453,
      "step": 26520
    },
    {
      "epoch": 0.21946478057658106,
      "grad_norm": 847.3948974609375,
      "learning_rate": 9.60671556010058e-06,
      "loss": 119.9163,
      "step": 26530
    },
    {
      "epoch": 0.21954750382595029,
      "grad_norm": 559.4891357421875,
      "learning_rate": 9.606167096083205e-06,
      "loss": 106.7365,
      "step": 26540
    },
    {
      "epoch": 0.2196302270753195,
      "grad_norm": 1348.5836181640625,
      "learning_rate": 9.60561826557425e-06,
      "loss": 148.4306,
      "step": 26550
    },
    {
      "epoch": 0.21971295032468877,
      "grad_norm": 912.4396362304688,
      "learning_rate": 9.60506906861739e-06,
      "loss": 116.1079,
      "step": 26560
    },
    {
      "epoch": 0.219795673574058,
      "grad_norm": 939.4193115234375,
      "learning_rate": 9.604519505256316e-06,
      "loss": 127.2096,
      "step": 26570
    },
    {
      "epoch": 0.21987839682342722,
      "grad_norm": 807.13623046875,
      "learning_rate": 9.603969575534757e-06,
      "loss": 102.2194,
      "step": 26580
    },
    {
      "epoch": 0.21996112007279647,
      "grad_norm": 695.100830078125,
      "learning_rate": 9.60341927949647e-06,
      "loss": 113.9714,
      "step": 26590
    },
    {
      "epoch": 0.2200438433221657,
      "grad_norm": 456.3763122558594,
      "learning_rate": 9.602868617185238e-06,
      "loss": 105.249,
      "step": 26600
    },
    {
      "epoch": 0.22012656657153493,
      "grad_norm": 1128.632568359375,
      "learning_rate": 9.602317588644872e-06,
      "loss": 104.7491,
      "step": 26610
    },
    {
      "epoch": 0.22020928982090415,
      "grad_norm": 948.9160766601562,
      "learning_rate": 9.601766193919217e-06,
      "loss": 104.4173,
      "step": 26620
    },
    {
      "epoch": 0.2202920130702734,
      "grad_norm": 725.5731201171875,
      "learning_rate": 9.601214433052147e-06,
      "loss": 103.2853,
      "step": 26630
    },
    {
      "epoch": 0.22037473631964263,
      "grad_norm": 979.1326293945312,
      "learning_rate": 9.600662306087562e-06,
      "loss": 122.0349,
      "step": 26640
    },
    {
      "epoch": 0.22045745956901186,
      "grad_norm": 772.8959350585938,
      "learning_rate": 9.600109813069389e-06,
      "loss": 118.9232,
      "step": 26650
    },
    {
      "epoch": 0.22054018281838111,
      "grad_norm": 879.559814453125,
      "learning_rate": 9.599556954041591e-06,
      "loss": 154.7716,
      "step": 26660
    },
    {
      "epoch": 0.22062290606775034,
      "grad_norm": 852.1553344726562,
      "learning_rate": 9.599003729048157e-06,
      "loss": 115.1464,
      "step": 26670
    },
    {
      "epoch": 0.22070562931711957,
      "grad_norm": 1091.1187744140625,
      "learning_rate": 9.598450138133101e-06,
      "loss": 124.5991,
      "step": 26680
    },
    {
      "epoch": 0.22078835256648882,
      "grad_norm": 1003.1347045898438,
      "learning_rate": 9.597896181340471e-06,
      "loss": 133.6112,
      "step": 26690
    },
    {
      "epoch": 0.22087107581585805,
      "grad_norm": 1310.1240234375,
      "learning_rate": 9.597341858714344e-06,
      "loss": 120.8151,
      "step": 26700
    },
    {
      "epoch": 0.22095379906522727,
      "grad_norm": 973.50439453125,
      "learning_rate": 9.596787170298824e-06,
      "loss": 132.3573,
      "step": 26710
    },
    {
      "epoch": 0.22103652231459653,
      "grad_norm": 838.1239013671875,
      "learning_rate": 9.596232116138047e-06,
      "loss": 135.2263,
      "step": 26720
    },
    {
      "epoch": 0.22111924556396576,
      "grad_norm": 776.5439453125,
      "learning_rate": 9.595676696276173e-06,
      "loss": 140.2891,
      "step": 26730
    },
    {
      "epoch": 0.22120196881333498,
      "grad_norm": 898.1226806640625,
      "learning_rate": 9.595120910757396e-06,
      "loss": 156.5662,
      "step": 26740
    },
    {
      "epoch": 0.22128469206270424,
      "grad_norm": 988.5482788085938,
      "learning_rate": 9.594564759625936e-06,
      "loss": 119.9368,
      "step": 26750
    },
    {
      "epoch": 0.22136741531207346,
      "grad_norm": 2739.469482421875,
      "learning_rate": 9.594008242926046e-06,
      "loss": 117.0178,
      "step": 26760
    },
    {
      "epoch": 0.2214501385614427,
      "grad_norm": 1172.0687255859375,
      "learning_rate": 9.593451360702003e-06,
      "loss": 109.8631,
      "step": 26770
    },
    {
      "epoch": 0.22153286181081192,
      "grad_norm": 1231.6884765625,
      "learning_rate": 9.592894112998115e-06,
      "loss": 123.6822,
      "step": 26780
    },
    {
      "epoch": 0.22161558506018117,
      "grad_norm": 2576.349609375,
      "learning_rate": 9.592336499858721e-06,
      "loss": 134.2115,
      "step": 26790
    },
    {
      "epoch": 0.2216983083095504,
      "grad_norm": 851.02197265625,
      "learning_rate": 9.59177852132819e-06,
      "loss": 130.6942,
      "step": 26800
    },
    {
      "epoch": 0.22178103155891962,
      "grad_norm": 1379.3516845703125,
      "learning_rate": 9.591220177450912e-06,
      "loss": 148.0982,
      "step": 26810
    },
    {
      "epoch": 0.22186375480828888,
      "grad_norm": 1033.928955078125,
      "learning_rate": 9.590661468271319e-06,
      "loss": 99.2162,
      "step": 26820
    },
    {
      "epoch": 0.2219464780576581,
      "grad_norm": 987.3099365234375,
      "learning_rate": 9.59010239383386e-06,
      "loss": 138.9152,
      "step": 26830
    },
    {
      "epoch": 0.22202920130702733,
      "grad_norm": 687.860595703125,
      "learning_rate": 9.589542954183018e-06,
      "loss": 112.7026,
      "step": 26840
    },
    {
      "epoch": 0.22211192455639658,
      "grad_norm": 732.00537109375,
      "learning_rate": 9.588983149363307e-06,
      "loss": 123.2144,
      "step": 26850
    },
    {
      "epoch": 0.2221946478057658,
      "grad_norm": 1079.510009765625,
      "learning_rate": 9.588422979419267e-06,
      "loss": 87.7841,
      "step": 26860
    },
    {
      "epoch": 0.22227737105513504,
      "grad_norm": 1468.6038818359375,
      "learning_rate": 9.587862444395471e-06,
      "loss": 136.3903,
      "step": 26870
    },
    {
      "epoch": 0.2223600943045043,
      "grad_norm": 889.5414428710938,
      "learning_rate": 9.587301544336513e-06,
      "loss": 115.5707,
      "step": 26880
    },
    {
      "epoch": 0.22244281755387352,
      "grad_norm": 590.138916015625,
      "learning_rate": 9.586740279287024e-06,
      "loss": 117.9152,
      "step": 26890
    },
    {
      "epoch": 0.22252554080324274,
      "grad_norm": 1332.5377197265625,
      "learning_rate": 9.586178649291664e-06,
      "loss": 131.7125,
      "step": 26900
    },
    {
      "epoch": 0.222608264052612,
      "grad_norm": 1022.7705688476562,
      "learning_rate": 9.585616654395113e-06,
      "loss": 115.9927,
      "step": 26910
    },
    {
      "epoch": 0.22269098730198122,
      "grad_norm": 847.87255859375,
      "learning_rate": 9.585054294642093e-06,
      "loss": 169.3321,
      "step": 26920
    },
    {
      "epoch": 0.22277371055135045,
      "grad_norm": 1865.4990234375,
      "learning_rate": 9.584491570077343e-06,
      "loss": 128.1739,
      "step": 26930
    },
    {
      "epoch": 0.2228564338007197,
      "grad_norm": 865.5852661132812,
      "learning_rate": 9.58392848074564e-06,
      "loss": 119.8757,
      "step": 26940
    },
    {
      "epoch": 0.22293915705008893,
      "grad_norm": 1116.7584228515625,
      "learning_rate": 9.583365026691785e-06,
      "loss": 111.9066,
      "step": 26950
    },
    {
      "epoch": 0.22302188029945816,
      "grad_norm": 1089.7393798828125,
      "learning_rate": 9.58280120796061e-06,
      "loss": 123.0073,
      "step": 26960
    },
    {
      "epoch": 0.22310460354882738,
      "grad_norm": 854.9454956054688,
      "learning_rate": 9.582237024596974e-06,
      "loss": 131.7255,
      "step": 26970
    },
    {
      "epoch": 0.22318732679819664,
      "grad_norm": 840.7254028320312,
      "learning_rate": 9.581672476645768e-06,
      "loss": 134.3853,
      "step": 26980
    },
    {
      "epoch": 0.22327005004756587,
      "grad_norm": 1240.8424072265625,
      "learning_rate": 9.58110756415191e-06,
      "loss": 127.8951,
      "step": 26990
    },
    {
      "epoch": 0.2233527732969351,
      "grad_norm": 1180.210205078125,
      "learning_rate": 9.580542287160348e-06,
      "loss": 127.6457,
      "step": 27000
    },
    {
      "epoch": 0.22343549654630435,
      "grad_norm": 1015.6724853515625,
      "learning_rate": 9.579976645716058e-06,
      "loss": 130.7046,
      "step": 27010
    },
    {
      "epoch": 0.22351821979567357,
      "grad_norm": 1194.459228515625,
      "learning_rate": 9.579410639864046e-06,
      "loss": 133.5198,
      "step": 27020
    },
    {
      "epoch": 0.2236009430450428,
      "grad_norm": 1310.887451171875,
      "learning_rate": 9.578844269649345e-06,
      "loss": 123.0892,
      "step": 27030
    },
    {
      "epoch": 0.22368366629441205,
      "grad_norm": 764.6407470703125,
      "learning_rate": 9.578277535117022e-06,
      "loss": 118.5598,
      "step": 27040
    },
    {
      "epoch": 0.22376638954378128,
      "grad_norm": 1107.6348876953125,
      "learning_rate": 9.577710436312164e-06,
      "loss": 113.8774,
      "step": 27050
    },
    {
      "epoch": 0.2238491127931505,
      "grad_norm": 1522.5843505859375,
      "learning_rate": 9.577142973279896e-06,
      "loss": 137.2552,
      "step": 27060
    },
    {
      "epoch": 0.22393183604251976,
      "grad_norm": 984.351318359375,
      "learning_rate": 9.576575146065369e-06,
      "loss": 128.6748,
      "step": 27070
    },
    {
      "epoch": 0.224014559291889,
      "grad_norm": 642.086181640625,
      "learning_rate": 9.576006954713762e-06,
      "loss": 117.2645,
      "step": 27080
    },
    {
      "epoch": 0.2240972825412582,
      "grad_norm": 1214.22998046875,
      "learning_rate": 9.57543839927028e-06,
      "loss": 128.8348,
      "step": 27090
    },
    {
      "epoch": 0.22418000579062747,
      "grad_norm": 1050.501953125,
      "learning_rate": 9.574869479780165e-06,
      "loss": 147.3516,
      "step": 27100
    },
    {
      "epoch": 0.2242627290399967,
      "grad_norm": 1185.0849609375,
      "learning_rate": 9.57430019628868e-06,
      "loss": 147.9033,
      "step": 27110
    },
    {
      "epoch": 0.22434545228936592,
      "grad_norm": 1115.831298828125,
      "learning_rate": 9.573730548841122e-06,
      "loss": 111.629,
      "step": 27120
    },
    {
      "epoch": 0.22442817553873518,
      "grad_norm": 1058.506103515625,
      "learning_rate": 9.573160537482816e-06,
      "loss": 135.1445,
      "step": 27130
    },
    {
      "epoch": 0.2245108987881044,
      "grad_norm": 1066.4493408203125,
      "learning_rate": 9.572590162259112e-06,
      "loss": 171.7336,
      "step": 27140
    },
    {
      "epoch": 0.22459362203747363,
      "grad_norm": 735.5531005859375,
      "learning_rate": 9.572019423215395e-06,
      "loss": 110.0656,
      "step": 27150
    },
    {
      "epoch": 0.22467634528684285,
      "grad_norm": 1075.277587890625,
      "learning_rate": 9.571448320397076e-06,
      "loss": 102.259,
      "step": 27160
    },
    {
      "epoch": 0.2247590685362121,
      "grad_norm": 677.2155151367188,
      "learning_rate": 9.570876853849593e-06,
      "loss": 128.2644,
      "step": 27170
    },
    {
      "epoch": 0.22484179178558134,
      "grad_norm": 993.1212768554688,
      "learning_rate": 9.570305023618417e-06,
      "loss": 185.3893,
      "step": 27180
    },
    {
      "epoch": 0.22492451503495056,
      "grad_norm": 963.5702514648438,
      "learning_rate": 9.569732829749045e-06,
      "loss": 125.3146,
      "step": 27190
    },
    {
      "epoch": 0.22500723828431982,
      "grad_norm": 950.5331420898438,
      "learning_rate": 9.569160272287003e-06,
      "loss": 126.4869,
      "step": 27200
    },
    {
      "epoch": 0.22508996153368904,
      "grad_norm": 820.460205078125,
      "learning_rate": 9.56858735127785e-06,
      "loss": 76.8966,
      "step": 27210
    },
    {
      "epoch": 0.22517268478305827,
      "grad_norm": 572.8544311523438,
      "learning_rate": 9.568014066767166e-06,
      "loss": 126.8829,
      "step": 27220
    },
    {
      "epoch": 0.22525540803242752,
      "grad_norm": 998.837646484375,
      "learning_rate": 9.567440418800569e-06,
      "loss": 134.7057,
      "step": 27230
    },
    {
      "epoch": 0.22533813128179675,
      "grad_norm": 956.578857421875,
      "learning_rate": 9.566866407423698e-06,
      "loss": 143.3908,
      "step": 27240
    },
    {
      "epoch": 0.22542085453116598,
      "grad_norm": 1049.07861328125,
      "learning_rate": 9.566292032682228e-06,
      "loss": 137.0985,
      "step": 27250
    },
    {
      "epoch": 0.22550357778053523,
      "grad_norm": 870.1298217773438,
      "learning_rate": 9.565717294621856e-06,
      "loss": 114.6946,
      "step": 27260
    },
    {
      "epoch": 0.22558630102990446,
      "grad_norm": 784.1261596679688,
      "learning_rate": 9.565142193288313e-06,
      "loss": 150.3023,
      "step": 27270
    },
    {
      "epoch": 0.22566902427927368,
      "grad_norm": 721.9214477539062,
      "learning_rate": 9.564566728727358e-06,
      "loss": 92.1085,
      "step": 27280
    },
    {
      "epoch": 0.22575174752864294,
      "grad_norm": 819.4923706054688,
      "learning_rate": 9.563990900984775e-06,
      "loss": 103.506,
      "step": 27290
    },
    {
      "epoch": 0.22583447077801216,
      "grad_norm": 2060.950927734375,
      "learning_rate": 9.563414710106382e-06,
      "loss": 212.3363,
      "step": 27300
    },
    {
      "epoch": 0.2259171940273814,
      "grad_norm": 921.5272216796875,
      "learning_rate": 9.562838156138025e-06,
      "loss": 142.4072,
      "step": 27310
    },
    {
      "epoch": 0.22599991727675062,
      "grad_norm": 935.8705444335938,
      "learning_rate": 9.562261239125575e-06,
      "loss": 122.9581,
      "step": 27320
    },
    {
      "epoch": 0.22608264052611987,
      "grad_norm": 925.080810546875,
      "learning_rate": 9.561683959114938e-06,
      "loss": 127.3665,
      "step": 27330
    },
    {
      "epoch": 0.2261653637754891,
      "grad_norm": 1195.8099365234375,
      "learning_rate": 9.561106316152043e-06,
      "loss": 133.1869,
      "step": 27340
    },
    {
      "epoch": 0.22624808702485832,
      "grad_norm": 784.7255249023438,
      "learning_rate": 9.56052831028285e-06,
      "loss": 139.8995,
      "step": 27350
    },
    {
      "epoch": 0.22633081027422758,
      "grad_norm": 1093.8494873046875,
      "learning_rate": 9.559949941553351e-06,
      "loss": 119.8402,
      "step": 27360
    },
    {
      "epoch": 0.2264135335235968,
      "grad_norm": 973.4978637695312,
      "learning_rate": 9.559371210009562e-06,
      "loss": 124.6592,
      "step": 27370
    },
    {
      "epoch": 0.22649625677296603,
      "grad_norm": 869.2335815429688,
      "learning_rate": 9.55879211569753e-06,
      "loss": 94.1615,
      "step": 27380
    },
    {
      "epoch": 0.22657898002233529,
      "grad_norm": 1302.50244140625,
      "learning_rate": 9.55821265866333e-06,
      "loss": 123.9871,
      "step": 27390
    },
    {
      "epoch": 0.2266617032717045,
      "grad_norm": 1908.8446044921875,
      "learning_rate": 9.55763283895307e-06,
      "loss": 140.2897,
      "step": 27400
    },
    {
      "epoch": 0.22674442652107374,
      "grad_norm": 553.2281494140625,
      "learning_rate": 9.557052656612882e-06,
      "loss": 98.0197,
      "step": 27410
    },
    {
      "epoch": 0.226827149770443,
      "grad_norm": 693.7083740234375,
      "learning_rate": 9.556472111688928e-06,
      "loss": 121.6574,
      "step": 27420
    },
    {
      "epoch": 0.22690987301981222,
      "grad_norm": 1643.4599609375,
      "learning_rate": 9.555891204227399e-06,
      "loss": 107.1005,
      "step": 27430
    },
    {
      "epoch": 0.22699259626918145,
      "grad_norm": 792.270263671875,
      "learning_rate": 9.555309934274515e-06,
      "loss": 177.7701,
      "step": 27440
    },
    {
      "epoch": 0.2270753195185507,
      "grad_norm": 456.06939697265625,
      "learning_rate": 9.554728301876525e-06,
      "loss": 137.5959,
      "step": 27450
    },
    {
      "epoch": 0.22715804276791993,
      "grad_norm": 966.9349365234375,
      "learning_rate": 9.554146307079711e-06,
      "loss": 99.2951,
      "step": 27460
    },
    {
      "epoch": 0.22724076601728915,
      "grad_norm": 1171.660400390625,
      "learning_rate": 9.553563949930374e-06,
      "loss": 126.8152,
      "step": 27470
    },
    {
      "epoch": 0.2273234892666584,
      "grad_norm": 631.3641967773438,
      "learning_rate": 9.552981230474849e-06,
      "loss": 101.0075,
      "step": 27480
    },
    {
      "epoch": 0.22740621251602763,
      "grad_norm": 1808.03466796875,
      "learning_rate": 9.552398148759506e-06,
      "loss": 105.2634,
      "step": 27490
    },
    {
      "epoch": 0.22748893576539686,
      "grad_norm": 1841.3206787109375,
      "learning_rate": 9.551814704830734e-06,
      "loss": 124.4455,
      "step": 27500
    },
    {
      "epoch": 0.2275716590147661,
      "grad_norm": 2790.470947265625,
      "learning_rate": 9.551230898734955e-06,
      "loss": 143.8618,
      "step": 27510
    },
    {
      "epoch": 0.22765438226413534,
      "grad_norm": 916.5533447265625,
      "learning_rate": 9.550646730518623e-06,
      "loss": 105.8261,
      "step": 27520
    },
    {
      "epoch": 0.22773710551350457,
      "grad_norm": 1437.979248046875,
      "learning_rate": 9.550062200228214e-06,
      "loss": 101.92,
      "step": 27530
    },
    {
      "epoch": 0.2278198287628738,
      "grad_norm": 1014.4559326171875,
      "learning_rate": 9.549477307910238e-06,
      "loss": 126.4266,
      "step": 27540
    },
    {
      "epoch": 0.22790255201224305,
      "grad_norm": 1683.48779296875,
      "learning_rate": 9.548892053611232e-06,
      "loss": 133.186,
      "step": 27550
    },
    {
      "epoch": 0.22798527526161227,
      "grad_norm": 864.87744140625,
      "learning_rate": 9.54830643737776e-06,
      "loss": 137.8783,
      "step": 27560
    },
    {
      "epoch": 0.2280679985109815,
      "grad_norm": 1702.0152587890625,
      "learning_rate": 9.54772045925642e-06,
      "loss": 125.6688,
      "step": 27570
    },
    {
      "epoch": 0.22815072176035076,
      "grad_norm": 1634.5037841796875,
      "learning_rate": 9.547134119293835e-06,
      "loss": 126.7895,
      "step": 27580
    },
    {
      "epoch": 0.22823344500971998,
      "grad_norm": 1126.988525390625,
      "learning_rate": 9.546547417536656e-06,
      "loss": 117.5014,
      "step": 27590
    },
    {
      "epoch": 0.2283161682590892,
      "grad_norm": 774.8274536132812,
      "learning_rate": 9.545960354031564e-06,
      "loss": 110.8326,
      "step": 27600
    },
    {
      "epoch": 0.22839889150845846,
      "grad_norm": 920.1649780273438,
      "learning_rate": 9.545372928825271e-06,
      "loss": 85.081,
      "step": 27610
    },
    {
      "epoch": 0.2284816147578277,
      "grad_norm": 550.8583374023438,
      "learning_rate": 9.544785141964514e-06,
      "loss": 82.1225,
      "step": 27620
    },
    {
      "epoch": 0.22856433800719692,
      "grad_norm": 2031.975830078125,
      "learning_rate": 9.544196993496062e-06,
      "loss": 113.8713,
      "step": 27630
    },
    {
      "epoch": 0.22864706125656617,
      "grad_norm": 660.8304443359375,
      "learning_rate": 9.54360848346671e-06,
      "loss": 111.5653,
      "step": 27640
    },
    {
      "epoch": 0.2287297845059354,
      "grad_norm": 1158.63916015625,
      "learning_rate": 9.543019611923283e-06,
      "loss": 105.64,
      "step": 27650
    },
    {
      "epoch": 0.22881250775530462,
      "grad_norm": 1066.19970703125,
      "learning_rate": 9.542430378912634e-06,
      "loss": 112.1879,
      "step": 27660
    },
    {
      "epoch": 0.22889523100467388,
      "grad_norm": 1097.3199462890625,
      "learning_rate": 9.541840784481648e-06,
      "loss": 112.1237,
      "step": 27670
    },
    {
      "epoch": 0.2289779542540431,
      "grad_norm": 1081.3756103515625,
      "learning_rate": 9.541250828677235e-06,
      "loss": 144.3419,
      "step": 27680
    },
    {
      "epoch": 0.22906067750341233,
      "grad_norm": 899.7421264648438,
      "learning_rate": 9.540660511546335e-06,
      "loss": 129.8028,
      "step": 27690
    },
    {
      "epoch": 0.22914340075278156,
      "grad_norm": 835.4717407226562,
      "learning_rate": 9.540069833135917e-06,
      "loss": 131.3196,
      "step": 27700
    },
    {
      "epoch": 0.2292261240021508,
      "grad_norm": 603.4605712890625,
      "learning_rate": 9.539478793492978e-06,
      "loss": 94.5269,
      "step": 27710
    },
    {
      "epoch": 0.22930884725152004,
      "grad_norm": 827.4793701171875,
      "learning_rate": 9.538887392664544e-06,
      "loss": 97.6406,
      "step": 27720
    },
    {
      "epoch": 0.22939157050088926,
      "grad_norm": 947.7173461914062,
      "learning_rate": 9.53829563069767e-06,
      "loss": 134.4241,
      "step": 27730
    },
    {
      "epoch": 0.22947429375025852,
      "grad_norm": 1094.2747802734375,
      "learning_rate": 9.537703507639444e-06,
      "loss": 102.8811,
      "step": 27740
    },
    {
      "epoch": 0.22955701699962774,
      "grad_norm": 1054.0640869140625,
      "learning_rate": 9.537111023536973e-06,
      "loss": 110.9093,
      "step": 27750
    },
    {
      "epoch": 0.22963974024899697,
      "grad_norm": 902.435302734375,
      "learning_rate": 9.536518178437402e-06,
      "loss": 120.4104,
      "step": 27760
    },
    {
      "epoch": 0.22972246349836623,
      "grad_norm": 1348.26025390625,
      "learning_rate": 9.535924972387898e-06,
      "loss": 109.3034,
      "step": 27770
    },
    {
      "epoch": 0.22980518674773545,
      "grad_norm": 857.2482299804688,
      "learning_rate": 9.535331405435662e-06,
      "loss": 125.6188,
      "step": 27780
    },
    {
      "epoch": 0.22988790999710468,
      "grad_norm": 1263.4981689453125,
      "learning_rate": 9.534737477627918e-06,
      "loss": 152.1994,
      "step": 27790
    },
    {
      "epoch": 0.22997063324647393,
      "grad_norm": 1043.9830322265625,
      "learning_rate": 9.534143189011928e-06,
      "loss": 139.1974,
      "step": 27800
    },
    {
      "epoch": 0.23005335649584316,
      "grad_norm": 1181.6470947265625,
      "learning_rate": 9.533548539634971e-06,
      "loss": 124.807,
      "step": 27810
    },
    {
      "epoch": 0.23013607974521239,
      "grad_norm": 1061.00244140625,
      "learning_rate": 9.532953529544365e-06,
      "loss": 127.3019,
      "step": 27820
    },
    {
      "epoch": 0.23021880299458164,
      "grad_norm": 2072.89404296875,
      "learning_rate": 9.532358158787446e-06,
      "loss": 112.2069,
      "step": 27830
    },
    {
      "epoch": 0.23030152624395087,
      "grad_norm": 726.18505859375,
      "learning_rate": 9.531762427411592e-06,
      "loss": 126.4197,
      "step": 27840
    },
    {
      "epoch": 0.2303842494933201,
      "grad_norm": 1991.278564453125,
      "learning_rate": 9.531166335464198e-06,
      "loss": 257.9386,
      "step": 27850
    },
    {
      "epoch": 0.23046697274268935,
      "grad_norm": 648.2035522460938,
      "learning_rate": 9.530569882992698e-06,
      "loss": 153.0686,
      "step": 27860
    },
    {
      "epoch": 0.23054969599205857,
      "grad_norm": 913.1694946289062,
      "learning_rate": 9.52997307004454e-06,
      "loss": 122.5074,
      "step": 27870
    },
    {
      "epoch": 0.2306324192414278,
      "grad_norm": 1065.9097900390625,
      "learning_rate": 9.529375896667218e-06,
      "loss": 95.8606,
      "step": 27880
    },
    {
      "epoch": 0.23071514249079703,
      "grad_norm": 2741.90234375,
      "learning_rate": 9.528778362908241e-06,
      "loss": 141.1961,
      "step": 27890
    },
    {
      "epoch": 0.23079786574016628,
      "grad_norm": 1098.0302734375,
      "learning_rate": 9.528180468815155e-06,
      "loss": 103.5171,
      "step": 27900
    },
    {
      "epoch": 0.2308805889895355,
      "grad_norm": 899.1497192382812,
      "learning_rate": 9.527582214435531e-06,
      "loss": 143.412,
      "step": 27910
    },
    {
      "epoch": 0.23096331223890473,
      "grad_norm": 311.1416320800781,
      "learning_rate": 9.526983599816968e-06,
      "loss": 91.3562,
      "step": 27920
    },
    {
      "epoch": 0.231046035488274,
      "grad_norm": 1649.770263671875,
      "learning_rate": 9.526384625007096e-06,
      "loss": 124.463,
      "step": 27930
    },
    {
      "epoch": 0.23112875873764321,
      "grad_norm": 804.8977661132812,
      "learning_rate": 9.525785290053573e-06,
      "loss": 138.7314,
      "step": 27940
    },
    {
      "epoch": 0.23121148198701244,
      "grad_norm": 859.3971557617188,
      "learning_rate": 9.525185595004085e-06,
      "loss": 85.416,
      "step": 27950
    },
    {
      "epoch": 0.2312942052363817,
      "grad_norm": 901.3175659179688,
      "learning_rate": 9.524585539906345e-06,
      "loss": 93.4797,
      "step": 27960
    },
    {
      "epoch": 0.23137692848575092,
      "grad_norm": 1204.8818359375,
      "learning_rate": 9.523985124808102e-06,
      "loss": 140.1107,
      "step": 27970
    },
    {
      "epoch": 0.23145965173512015,
      "grad_norm": 1186.2174072265625,
      "learning_rate": 9.523384349757123e-06,
      "loss": 111.8204,
      "step": 27980
    },
    {
      "epoch": 0.2315423749844894,
      "grad_norm": 1178.4058837890625,
      "learning_rate": 9.522783214801213e-06,
      "loss": 164.7376,
      "step": 27990
    },
    {
      "epoch": 0.23162509823385863,
      "grad_norm": 864.9526977539062,
      "learning_rate": 9.522181719988196e-06,
      "loss": 112.4256,
      "step": 28000
    },
    {
      "epoch": 0.23170782148322785,
      "grad_norm": 1473.5499267578125,
      "learning_rate": 9.521579865365935e-06,
      "loss": 117.2831,
      "step": 28010
    },
    {
      "epoch": 0.2317905447325971,
      "grad_norm": 525.5335083007812,
      "learning_rate": 9.520977650982316e-06,
      "loss": 150.6173,
      "step": 28020
    },
    {
      "epoch": 0.23187326798196634,
      "grad_norm": 886.9190063476562,
      "learning_rate": 9.520375076885253e-06,
      "loss": 146.7385,
      "step": 28030
    },
    {
      "epoch": 0.23195599123133556,
      "grad_norm": 937.6173095703125,
      "learning_rate": 9.519772143122691e-06,
      "loss": 104.6981,
      "step": 28040
    },
    {
      "epoch": 0.2320387144807048,
      "grad_norm": 834.8984375,
      "learning_rate": 9.519168849742603e-06,
      "loss": 96.541,
      "step": 28050
    },
    {
      "epoch": 0.23212143773007404,
      "grad_norm": 863.10986328125,
      "learning_rate": 9.51856519679299e-06,
      "loss": 93.61,
      "step": 28060
    },
    {
      "epoch": 0.23220416097944327,
      "grad_norm": 683.4520263671875,
      "learning_rate": 9.517961184321882e-06,
      "loss": 131.8342,
      "step": 28070
    },
    {
      "epoch": 0.2322868842288125,
      "grad_norm": 2610.283203125,
      "learning_rate": 9.517356812377336e-06,
      "loss": 189.3643,
      "step": 28080
    },
    {
      "epoch": 0.23236960747818175,
      "grad_norm": 993.5272827148438,
      "learning_rate": 9.516752081007441e-06,
      "loss": 133.2566,
      "step": 28090
    },
    {
      "epoch": 0.23245233072755098,
      "grad_norm": 1212.4417724609375,
      "learning_rate": 9.51614699026031e-06,
      "loss": 129.3865,
      "step": 28100
    },
    {
      "epoch": 0.2325350539769202,
      "grad_norm": 812.443359375,
      "learning_rate": 9.515541540184093e-06,
      "loss": 95.0065,
      "step": 28110
    },
    {
      "epoch": 0.23261777722628946,
      "grad_norm": 1204.5474853515625,
      "learning_rate": 9.514935730826957e-06,
      "loss": 145.4519,
      "step": 28120
    },
    {
      "epoch": 0.23270050047565868,
      "grad_norm": 1555.590576171875,
      "learning_rate": 9.514329562237107e-06,
      "loss": 136.7933,
      "step": 28130
    },
    {
      "epoch": 0.2327832237250279,
      "grad_norm": 870.7056274414062,
      "learning_rate": 9.51372303446277e-06,
      "loss": 122.1039,
      "step": 28140
    },
    {
      "epoch": 0.23286594697439716,
      "grad_norm": 1071.2955322265625,
      "learning_rate": 9.513116147552207e-06,
      "loss": 102.6043,
      "step": 28150
    },
    {
      "epoch": 0.2329486702237664,
      "grad_norm": 797.454833984375,
      "learning_rate": 9.512508901553703e-06,
      "loss": 140.6481,
      "step": 28160
    },
    {
      "epoch": 0.23303139347313562,
      "grad_norm": 585.2443237304688,
      "learning_rate": 9.511901296515578e-06,
      "loss": 113.4713,
      "step": 28170
    },
    {
      "epoch": 0.23311411672250487,
      "grad_norm": 870.2528076171875,
      "learning_rate": 9.511293332486172e-06,
      "loss": 130.058,
      "step": 28180
    },
    {
      "epoch": 0.2331968399718741,
      "grad_norm": 1487.830322265625,
      "learning_rate": 9.51068500951386e-06,
      "loss": 132.6228,
      "step": 28190
    },
    {
      "epoch": 0.23327956322124332,
      "grad_norm": 3511.711181640625,
      "learning_rate": 9.510076327647043e-06,
      "loss": 113.2403,
      "step": 28200
    },
    {
      "epoch": 0.23336228647061258,
      "grad_norm": 1605.590087890625,
      "learning_rate": 9.509467286934151e-06,
      "loss": 150.1105,
      "step": 28210
    },
    {
      "epoch": 0.2334450097199818,
      "grad_norm": 1058.44189453125,
      "learning_rate": 9.508857887423644e-06,
      "loss": 122.3643,
      "step": 28220
    },
    {
      "epoch": 0.23352773296935103,
      "grad_norm": 820.7523803710938,
      "learning_rate": 9.508248129164006e-06,
      "loss": 80.4105,
      "step": 28230
    },
    {
      "epoch": 0.23361045621872026,
      "grad_norm": 912.558837890625,
      "learning_rate": 9.507638012203755e-06,
      "loss": 152.501,
      "step": 28240
    },
    {
      "epoch": 0.2336931794680895,
      "grad_norm": 1337.6898193359375,
      "learning_rate": 9.507027536591436e-06,
      "loss": 149.5806,
      "step": 28250
    },
    {
      "epoch": 0.23377590271745874,
      "grad_norm": 1526.1043701171875,
      "learning_rate": 9.506416702375618e-06,
      "loss": 153.4466,
      "step": 28260
    },
    {
      "epoch": 0.23385862596682797,
      "grad_norm": 933.4179077148438,
      "learning_rate": 9.505805509604906e-06,
      "loss": 106.969,
      "step": 28270
    },
    {
      "epoch": 0.23394134921619722,
      "grad_norm": 1189.154296875,
      "learning_rate": 9.505193958327927e-06,
      "loss": 129.8097,
      "step": 28280
    },
    {
      "epoch": 0.23402407246556645,
      "grad_norm": 691.6627197265625,
      "learning_rate": 9.504582048593343e-06,
      "loss": 99.3678,
      "step": 28290
    },
    {
      "epoch": 0.23410679571493567,
      "grad_norm": 1236.778076171875,
      "learning_rate": 9.503969780449838e-06,
      "loss": 119.0243,
      "step": 28300
    },
    {
      "epoch": 0.23418951896430493,
      "grad_norm": 1243.570068359375,
      "learning_rate": 9.503357153946126e-06,
      "loss": 104.1002,
      "step": 28310
    },
    {
      "epoch": 0.23427224221367415,
      "grad_norm": 1092.5941162109375,
      "learning_rate": 9.502744169130955e-06,
      "loss": 97.2079,
      "step": 28320
    },
    {
      "epoch": 0.23435496546304338,
      "grad_norm": 1051.38671875,
      "learning_rate": 9.502130826053095e-06,
      "loss": 132.0031,
      "step": 28330
    },
    {
      "epoch": 0.23443768871241263,
      "grad_norm": 1343.4345703125,
      "learning_rate": 9.501517124761347e-06,
      "loss": 112.9695,
      "step": 28340
    },
    {
      "epoch": 0.23452041196178186,
      "grad_norm": 637.9159545898438,
      "learning_rate": 9.50090306530454e-06,
      "loss": 85.6707,
      "step": 28350
    },
    {
      "epoch": 0.2346031352111511,
      "grad_norm": 976.9581909179688,
      "learning_rate": 9.500288647731533e-06,
      "loss": 127.3839,
      "step": 28360
    },
    {
      "epoch": 0.23468585846052034,
      "grad_norm": 1000.0506591796875,
      "learning_rate": 9.49967387209121e-06,
      "loss": 168.1098,
      "step": 28370
    },
    {
      "epoch": 0.23476858170988957,
      "grad_norm": 838.3327026367188,
      "learning_rate": 9.499058738432492e-06,
      "loss": 112.4135,
      "step": 28380
    },
    {
      "epoch": 0.2348513049592588,
      "grad_norm": 708.0794067382812,
      "learning_rate": 9.498443246804314e-06,
      "loss": 120.7116,
      "step": 28390
    },
    {
      "epoch": 0.23493402820862805,
      "grad_norm": 1063.5574951171875,
      "learning_rate": 9.497827397255655e-06,
      "loss": 99.907,
      "step": 28400
    },
    {
      "epoch": 0.23501675145799727,
      "grad_norm": 933.8026123046875,
      "learning_rate": 9.49721118983551e-06,
      "loss": 153.6268,
      "step": 28410
    },
    {
      "epoch": 0.2350994747073665,
      "grad_norm": 836.0083618164062,
      "learning_rate": 9.49659462459291e-06,
      "loss": 123.5795,
      "step": 28420
    },
    {
      "epoch": 0.23518219795673573,
      "grad_norm": 1091.86669921875,
      "learning_rate": 9.495977701576913e-06,
      "loss": 141.2293,
      "step": 28430
    },
    {
      "epoch": 0.23526492120610498,
      "grad_norm": 960.7584838867188,
      "learning_rate": 9.495360420836603e-06,
      "loss": 113.6143,
      "step": 28440
    },
    {
      "epoch": 0.2353476444554742,
      "grad_norm": 1572.40771484375,
      "learning_rate": 9.494742782421099e-06,
      "loss": 159.3734,
      "step": 28450
    },
    {
      "epoch": 0.23543036770484344,
      "grad_norm": 1193.2608642578125,
      "learning_rate": 9.494124786379535e-06,
      "loss": 128.3347,
      "step": 28460
    },
    {
      "epoch": 0.2355130909542127,
      "grad_norm": 1346.21630859375,
      "learning_rate": 9.49350643276109e-06,
      "loss": 105.2375,
      "step": 28470
    },
    {
      "epoch": 0.23559581420358192,
      "grad_norm": 1731.52001953125,
      "learning_rate": 9.49288772161496e-06,
      "loss": 134.3214,
      "step": 28480
    },
    {
      "epoch": 0.23567853745295114,
      "grad_norm": 982.36279296875,
      "learning_rate": 9.492268652990374e-06,
      "loss": 120.5295,
      "step": 28490
    },
    {
      "epoch": 0.2357612607023204,
      "grad_norm": 1203.948486328125,
      "learning_rate": 9.491649226936586e-06,
      "loss": 149.1518,
      "step": 28500
    },
    {
      "epoch": 0.23584398395168962,
      "grad_norm": 1445.0263671875,
      "learning_rate": 9.491029443502884e-06,
      "loss": 133.0336,
      "step": 28510
    },
    {
      "epoch": 0.23592670720105885,
      "grad_norm": 1120.74462890625,
      "learning_rate": 9.490409302738582e-06,
      "loss": 104.1838,
      "step": 28520
    },
    {
      "epoch": 0.2360094304504281,
      "grad_norm": 1055.50830078125,
      "learning_rate": 9.489788804693017e-06,
      "loss": 97.1542,
      "step": 28530
    },
    {
      "epoch": 0.23609215369979733,
      "grad_norm": 645.5867309570312,
      "learning_rate": 9.489167949415563e-06,
      "loss": 124.6525,
      "step": 28540
    },
    {
      "epoch": 0.23617487694916656,
      "grad_norm": 723.2225952148438,
      "learning_rate": 9.48854673695562e-06,
      "loss": 117.6365,
      "step": 28550
    },
    {
      "epoch": 0.2362576001985358,
      "grad_norm": 1657.0540771484375,
      "learning_rate": 9.48792516736261e-06,
      "loss": 131.2895,
      "step": 28560
    },
    {
      "epoch": 0.23634032344790504,
      "grad_norm": 967.8582763671875,
      "learning_rate": 9.487303240685992e-06,
      "loss": 100.9019,
      "step": 28570
    },
    {
      "epoch": 0.23642304669727426,
      "grad_norm": 2011.5921630859375,
      "learning_rate": 9.48668095697525e-06,
      "loss": 167.2856,
      "step": 28580
    },
    {
      "epoch": 0.2365057699466435,
      "grad_norm": 991.98291015625,
      "learning_rate": 9.486058316279894e-06,
      "loss": 158.8021,
      "step": 28590
    },
    {
      "epoch": 0.23658849319601274,
      "grad_norm": 905.6279296875,
      "learning_rate": 9.485435318649468e-06,
      "loss": 124.5288,
      "step": 28600
    },
    {
      "epoch": 0.23667121644538197,
      "grad_norm": 793.5623779296875,
      "learning_rate": 9.484811964133537e-06,
      "loss": 138.7439,
      "step": 28610
    },
    {
      "epoch": 0.2367539396947512,
      "grad_norm": 1176.435791015625,
      "learning_rate": 9.484188252781701e-06,
      "loss": 109.5845,
      "step": 28620
    },
    {
      "epoch": 0.23683666294412045,
      "grad_norm": 840.8705444335938,
      "learning_rate": 9.483564184643586e-06,
      "loss": 90.2001,
      "step": 28630
    },
    {
      "epoch": 0.23691938619348968,
      "grad_norm": 705.6690673828125,
      "learning_rate": 9.482939759768845e-06,
      "loss": 145.6554,
      "step": 28640
    },
    {
      "epoch": 0.2370021094428589,
      "grad_norm": 1252.253662109375,
      "learning_rate": 9.48231497820716e-06,
      "loss": 129.5605,
      "step": 28650
    },
    {
      "epoch": 0.23708483269222816,
      "grad_norm": 1174.8831787109375,
      "learning_rate": 9.481689840008246e-06,
      "loss": 121.5843,
      "step": 28660
    },
    {
      "epoch": 0.23716755594159739,
      "grad_norm": 1164.958740234375,
      "learning_rate": 9.481064345221838e-06,
      "loss": 130.8124,
      "step": 28670
    },
    {
      "epoch": 0.2372502791909666,
      "grad_norm": 1460.279052734375,
      "learning_rate": 9.480438493897707e-06,
      "loss": 186.2501,
      "step": 28680
    },
    {
      "epoch": 0.23733300244033587,
      "grad_norm": 1548.7734375,
      "learning_rate": 9.479812286085645e-06,
      "loss": 122.9342,
      "step": 28690
    },
    {
      "epoch": 0.2374157256897051,
      "grad_norm": 485.3841857910156,
      "learning_rate": 9.47918572183548e-06,
      "loss": 116.2105,
      "step": 28700
    },
    {
      "epoch": 0.23749844893907432,
      "grad_norm": 2227.711181640625,
      "learning_rate": 9.478558801197065e-06,
      "loss": 108.795,
      "step": 28710
    },
    {
      "epoch": 0.23758117218844357,
      "grad_norm": 1480.4208984375,
      "learning_rate": 9.47793152422028e-06,
      "loss": 103.7922,
      "step": 28720
    },
    {
      "epoch": 0.2376638954378128,
      "grad_norm": 845.8829956054688,
      "learning_rate": 9.477303890955032e-06,
      "loss": 112.599,
      "step": 28730
    },
    {
      "epoch": 0.23774661868718203,
      "grad_norm": 593.84619140625,
      "learning_rate": 9.476675901451264e-06,
      "loss": 124.1586,
      "step": 28740
    },
    {
      "epoch": 0.23782934193655128,
      "grad_norm": 636.6671752929688,
      "learning_rate": 9.476047555758938e-06,
      "loss": 172.5131,
      "step": 28750
    },
    {
      "epoch": 0.2379120651859205,
      "grad_norm": 4360.0341796875,
      "learning_rate": 9.475418853928051e-06,
      "loss": 191.1747,
      "step": 28760
    },
    {
      "epoch": 0.23799478843528973,
      "grad_norm": 483.2803649902344,
      "learning_rate": 9.474789796008625e-06,
      "loss": 138.2722,
      "step": 28770
    },
    {
      "epoch": 0.23807751168465896,
      "grad_norm": 1149.00048828125,
      "learning_rate": 9.474160382050711e-06,
      "loss": 126.3032,
      "step": 28780
    },
    {
      "epoch": 0.23816023493402821,
      "grad_norm": 1440.6688232421875,
      "learning_rate": 9.47353061210439e-06,
      "loss": 96.6904,
      "step": 28790
    },
    {
      "epoch": 0.23824295818339744,
      "grad_norm": 2017.08837890625,
      "learning_rate": 9.47290048621977e-06,
      "loss": 116.2174,
      "step": 28800
    },
    {
      "epoch": 0.23832568143276667,
      "grad_norm": 900.5809326171875,
      "learning_rate": 9.472270004446984e-06,
      "loss": 110.8572,
      "step": 28810
    },
    {
      "epoch": 0.23840840468213592,
      "grad_norm": 853.006591796875,
      "learning_rate": 9.4716391668362e-06,
      "loss": 115.7237,
      "step": 28820
    },
    {
      "epoch": 0.23849112793150515,
      "grad_norm": 1115.9259033203125,
      "learning_rate": 9.471007973437607e-06,
      "loss": 108.8435,
      "step": 28830
    },
    {
      "epoch": 0.23857385118087437,
      "grad_norm": 1762.0460205078125,
      "learning_rate": 9.470376424301432e-06,
      "loss": 148.8191,
      "step": 28840
    },
    {
      "epoch": 0.23865657443024363,
      "grad_norm": 1468.8228759765625,
      "learning_rate": 9.46974451947792e-06,
      "loss": 154.2129,
      "step": 28850
    },
    {
      "epoch": 0.23873929767961286,
      "grad_norm": 1209.044921875,
      "learning_rate": 9.469112259017349e-06,
      "loss": 107.4766,
      "step": 28860
    },
    {
      "epoch": 0.23882202092898208,
      "grad_norm": 1850.443603515625,
      "learning_rate": 9.468479642970027e-06,
      "loss": 117.6253,
      "step": 28870
    },
    {
      "epoch": 0.23890474417835134,
      "grad_norm": 1324.356689453125,
      "learning_rate": 9.467846671386287e-06,
      "loss": 178.3749,
      "step": 28880
    },
    {
      "epoch": 0.23898746742772056,
      "grad_norm": 1060.7711181640625,
      "learning_rate": 9.467213344316493e-06,
      "loss": 101.0151,
      "step": 28890
    },
    {
      "epoch": 0.2390701906770898,
      "grad_norm": 2596.78076171875,
      "learning_rate": 9.466579661811032e-06,
      "loss": 149.5662,
      "step": 28900
    },
    {
      "epoch": 0.23915291392645904,
      "grad_norm": 1160.0654296875,
      "learning_rate": 9.46594562392033e-06,
      "loss": 114.8278,
      "step": 28910
    },
    {
      "epoch": 0.23923563717582827,
      "grad_norm": 1474.780029296875,
      "learning_rate": 9.465311230694828e-06,
      "loss": 94.9893,
      "step": 28920
    },
    {
      "epoch": 0.2393183604251975,
      "grad_norm": 827.9722900390625,
      "learning_rate": 9.464676482185005e-06,
      "loss": 82.4494,
      "step": 28930
    },
    {
      "epoch": 0.23940108367456675,
      "grad_norm": 490.88360595703125,
      "learning_rate": 9.464041378441365e-06,
      "loss": 160.122,
      "step": 28940
    },
    {
      "epoch": 0.23948380692393598,
      "grad_norm": 2539.133544921875,
      "learning_rate": 9.46340591951444e-06,
      "loss": 135.2263,
      "step": 28950
    },
    {
      "epoch": 0.2395665301733052,
      "grad_norm": 620.4109497070312,
      "learning_rate": 9.462770105454789e-06,
      "loss": 130.8018,
      "step": 28960
    },
    {
      "epoch": 0.23964925342267443,
      "grad_norm": 575.00146484375,
      "learning_rate": 9.462133936313002e-06,
      "loss": 123.9707,
      "step": 28970
    },
    {
      "epoch": 0.23973197667204368,
      "grad_norm": 764.5715942382812,
      "learning_rate": 9.461497412139697e-06,
      "loss": 103.3378,
      "step": 28980
    },
    {
      "epoch": 0.2398146999214129,
      "grad_norm": 1179.3966064453125,
      "learning_rate": 9.46086053298552e-06,
      "loss": 139.9406,
      "step": 28990
    },
    {
      "epoch": 0.23989742317078214,
      "grad_norm": 1064.7276611328125,
      "learning_rate": 9.460223298901138e-06,
      "loss": 100.375,
      "step": 29000
    },
    {
      "epoch": 0.2399801464201514,
      "grad_norm": 833.2089233398438,
      "learning_rate": 9.459585709937262e-06,
      "loss": 120.6056,
      "step": 29010
    },
    {
      "epoch": 0.24006286966952062,
      "grad_norm": 1361.6549072265625,
      "learning_rate": 9.458947766144617e-06,
      "loss": 129.4685,
      "step": 29020
    },
    {
      "epoch": 0.24014559291888984,
      "grad_norm": 1131.559326171875,
      "learning_rate": 9.458309467573963e-06,
      "loss": 90.7656,
      "step": 29030
    },
    {
      "epoch": 0.2402283161682591,
      "grad_norm": 1216.7392578125,
      "learning_rate": 9.457670814276083e-06,
      "loss": 105.4316,
      "step": 29040
    },
    {
      "epoch": 0.24031103941762832,
      "grad_norm": 853.8714599609375,
      "learning_rate": 9.457031806301795e-06,
      "loss": 94.2898,
      "step": 29050
    },
    {
      "epoch": 0.24039376266699755,
      "grad_norm": 826.33837890625,
      "learning_rate": 9.456392443701943e-06,
      "loss": 118.5048,
      "step": 29060
    },
    {
      "epoch": 0.2404764859163668,
      "grad_norm": 969.9588012695312,
      "learning_rate": 9.455752726527395e-06,
      "loss": 158.1088,
      "step": 29070
    },
    {
      "epoch": 0.24055920916573603,
      "grad_norm": 591.8674926757812,
      "learning_rate": 9.45511265482905e-06,
      "loss": 106.0139,
      "step": 29080
    },
    {
      "epoch": 0.24064193241510526,
      "grad_norm": 1056.0914306640625,
      "learning_rate": 9.454472228657841e-06,
      "loss": 148.1635,
      "step": 29090
    },
    {
      "epoch": 0.2407246556644745,
      "grad_norm": 1000.041259765625,
      "learning_rate": 9.453831448064717e-06,
      "loss": 119.1304,
      "step": 29100
    },
    {
      "epoch": 0.24080737891384374,
      "grad_norm": 765.8133544921875,
      "learning_rate": 9.453190313100666e-06,
      "loss": 83.0749,
      "step": 29110
    },
    {
      "epoch": 0.24089010216321297,
      "grad_norm": 963.5242919921875,
      "learning_rate": 9.4525488238167e-06,
      "loss": 153.7064,
      "step": 29120
    },
    {
      "epoch": 0.24097282541258222,
      "grad_norm": 809.49267578125,
      "learning_rate": 9.451906980263857e-06,
      "loss": 122.1319,
      "step": 29130
    },
    {
      "epoch": 0.24105554866195145,
      "grad_norm": 800.9232788085938,
      "learning_rate": 9.451264782493208e-06,
      "loss": 101.5012,
      "step": 29140
    },
    {
      "epoch": 0.24113827191132067,
      "grad_norm": 1233.4398193359375,
      "learning_rate": 9.450622230555849e-06,
      "loss": 144.6246,
      "step": 29150
    },
    {
      "epoch": 0.2412209951606899,
      "grad_norm": 692.1824951171875,
      "learning_rate": 9.449979324502905e-06,
      "loss": 160.0062,
      "step": 29160
    },
    {
      "epoch": 0.24130371841005915,
      "grad_norm": 946.5017700195312,
      "learning_rate": 9.449336064385529e-06,
      "loss": 105.0953,
      "step": 29170
    },
    {
      "epoch": 0.24138644165942838,
      "grad_norm": 1230.076416015625,
      "learning_rate": 9.4486924502549e-06,
      "loss": 120.0082,
      "step": 29180
    },
    {
      "epoch": 0.2414691649087976,
      "grad_norm": 839.8562622070312,
      "learning_rate": 9.448048482162231e-06,
      "loss": 137.8697,
      "step": 29190
    },
    {
      "epoch": 0.24155188815816686,
      "grad_norm": 882.9497680664062,
      "learning_rate": 9.447404160158758e-06,
      "loss": 119.5869,
      "step": 29200
    },
    {
      "epoch": 0.2416346114075361,
      "grad_norm": 1333.1221923828125,
      "learning_rate": 9.446759484295745e-06,
      "loss": 116.6337,
      "step": 29210
    },
    {
      "epoch": 0.2417173346569053,
      "grad_norm": 955.5294799804688,
      "learning_rate": 9.44611445462449e-06,
      "loss": 134.2271,
      "step": 29220
    },
    {
      "epoch": 0.24180005790627457,
      "grad_norm": 1251.0631103515625,
      "learning_rate": 9.445469071196312e-06,
      "loss": 124.4641,
      "step": 29230
    },
    {
      "epoch": 0.2418827811556438,
      "grad_norm": 834.0491943359375,
      "learning_rate": 9.444823334062562e-06,
      "loss": 116.2968,
      "step": 29240
    },
    {
      "epoch": 0.24196550440501302,
      "grad_norm": 1349.356201171875,
      "learning_rate": 9.444177243274619e-06,
      "loss": 131.4607,
      "step": 29250
    },
    {
      "epoch": 0.24204822765438228,
      "grad_norm": 1118.03173828125,
      "learning_rate": 9.443530798883887e-06,
      "loss": 107.2266,
      "step": 29260
    },
    {
      "epoch": 0.2421309509037515,
      "grad_norm": 1354.8619384765625,
      "learning_rate": 9.442884000941803e-06,
      "loss": 129.6626,
      "step": 29270
    },
    {
      "epoch": 0.24221367415312073,
      "grad_norm": 1301.5723876953125,
      "learning_rate": 9.44223684949983e-06,
      "loss": 126.9563,
      "step": 29280
    },
    {
      "epoch": 0.24229639740248998,
      "grad_norm": 639.9192504882812,
      "learning_rate": 9.441589344609457e-06,
      "loss": 97.4439,
      "step": 29290
    },
    {
      "epoch": 0.2423791206518592,
      "grad_norm": 1386.3795166015625,
      "learning_rate": 9.440941486322205e-06,
      "loss": 150.2773,
      "step": 29300
    },
    {
      "epoch": 0.24246184390122844,
      "grad_norm": 1358.3714599609375,
      "learning_rate": 9.44029327468962e-06,
      "loss": 110.78,
      "step": 29310
    },
    {
      "epoch": 0.24254456715059766,
      "grad_norm": 757.8045043945312,
      "learning_rate": 9.439644709763276e-06,
      "loss": 112.67,
      "step": 29320
    },
    {
      "epoch": 0.24262729039996692,
      "grad_norm": 930.5925903320312,
      "learning_rate": 9.43899579159478e-06,
      "loss": 121.9743,
      "step": 29330
    },
    {
      "epoch": 0.24271001364933614,
      "grad_norm": 1049.4073486328125,
      "learning_rate": 9.438346520235759e-06,
      "loss": 100.5406,
      "step": 29340
    },
    {
      "epoch": 0.24279273689870537,
      "grad_norm": 1115.40966796875,
      "learning_rate": 9.437696895737876e-06,
      "loss": 121.6903,
      "step": 29350
    },
    {
      "epoch": 0.24287546014807462,
      "grad_norm": 698.0397338867188,
      "learning_rate": 9.437046918152817e-06,
      "loss": 88.6896,
      "step": 29360
    },
    {
      "epoch": 0.24295818339744385,
      "grad_norm": 824.7769165039062,
      "learning_rate": 9.436396587532297e-06,
      "loss": 126.8226,
      "step": 29370
    },
    {
      "epoch": 0.24304090664681308,
      "grad_norm": 1229.65869140625,
      "learning_rate": 9.435745903928062e-06,
      "loss": 113.3302,
      "step": 29380
    },
    {
      "epoch": 0.24312362989618233,
      "grad_norm": 845.088623046875,
      "learning_rate": 9.435094867391881e-06,
      "loss": 154.009,
      "step": 29390
    },
    {
      "epoch": 0.24320635314555156,
      "grad_norm": 961.1011962890625,
      "learning_rate": 9.434443477975557e-06,
      "loss": 103.9956,
      "step": 29400
    },
    {
      "epoch": 0.24328907639492078,
      "grad_norm": 757.7135009765625,
      "learning_rate": 9.433791735730917e-06,
      "loss": 98.1805,
      "step": 29410
    },
    {
      "epoch": 0.24337179964429004,
      "grad_norm": 1421.3348388671875,
      "learning_rate": 9.433139640709817e-06,
      "loss": 132.9433,
      "step": 29420
    },
    {
      "epoch": 0.24345452289365926,
      "grad_norm": 608.5304565429688,
      "learning_rate": 9.432487192964142e-06,
      "loss": 122.0067,
      "step": 29430
    },
    {
      "epoch": 0.2435372461430285,
      "grad_norm": 1166.598876953125,
      "learning_rate": 9.431834392545803e-06,
      "loss": 127.8436,
      "step": 29440
    },
    {
      "epoch": 0.24361996939239774,
      "grad_norm": 1254.7440185546875,
      "learning_rate": 9.43118123950674e-06,
      "loss": 124.5958,
      "step": 29450
    },
    {
      "epoch": 0.24370269264176697,
      "grad_norm": 924.8063354492188,
      "learning_rate": 9.430527733898922e-06,
      "loss": 102.3073,
      "step": 29460
    },
    {
      "epoch": 0.2437854158911362,
      "grad_norm": 957.7825927734375,
      "learning_rate": 9.429873875774344e-06,
      "loss": 112.2574,
      "step": 29470
    },
    {
      "epoch": 0.24386813914050545,
      "grad_norm": 867.5020141601562,
      "learning_rate": 9.429219665185034e-06,
      "loss": 109.0799,
      "step": 29480
    },
    {
      "epoch": 0.24395086238987468,
      "grad_norm": 1606.21435546875,
      "learning_rate": 9.428565102183043e-06,
      "loss": 114.3639,
      "step": 29490
    },
    {
      "epoch": 0.2440335856392439,
      "grad_norm": 782.0588989257812,
      "learning_rate": 9.42791018682045e-06,
      "loss": 102.8368,
      "step": 29500
    },
    {
      "epoch": 0.24411630888861313,
      "grad_norm": 1499.0159912109375,
      "learning_rate": 9.427254919149367e-06,
      "loss": 129.8493,
      "step": 29510
    },
    {
      "epoch": 0.24419903213798239,
      "grad_norm": 1734.2025146484375,
      "learning_rate": 9.426599299221925e-06,
      "loss": 118.0028,
      "step": 29520
    },
    {
      "epoch": 0.2442817553873516,
      "grad_norm": 1104.4652099609375,
      "learning_rate": 9.425943327090295e-06,
      "loss": 133.7769,
      "step": 29530
    },
    {
      "epoch": 0.24436447863672084,
      "grad_norm": 809.5218505859375,
      "learning_rate": 9.425287002806666e-06,
      "loss": 101.5154,
      "step": 29540
    },
    {
      "epoch": 0.2444472018860901,
      "grad_norm": 860.76123046875,
      "learning_rate": 9.42463032642326e-06,
      "loss": 126.5965,
      "step": 29550
    },
    {
      "epoch": 0.24452992513545932,
      "grad_norm": 1226.3048095703125,
      "learning_rate": 9.423973297992324e-06,
      "loss": 133.2678,
      "step": 29560
    },
    {
      "epoch": 0.24461264838482855,
      "grad_norm": 638.6930541992188,
      "learning_rate": 9.423315917566137e-06,
      "loss": 153.0996,
      "step": 29570
    },
    {
      "epoch": 0.2446953716341978,
      "grad_norm": 598.7249755859375,
      "learning_rate": 9.422658185197002e-06,
      "loss": 122.7943,
      "step": 29580
    },
    {
      "epoch": 0.24477809488356703,
      "grad_norm": 711.6947021484375,
      "learning_rate": 9.422000100937253e-06,
      "loss": 93.9475,
      "step": 29590
    },
    {
      "epoch": 0.24486081813293625,
      "grad_norm": 1122.1689453125,
      "learning_rate": 9.42134166483925e-06,
      "loss": 97.7506,
      "step": 29600
    },
    {
      "epoch": 0.2449435413823055,
      "grad_norm": 1915.1302490234375,
      "learning_rate": 9.420682876955382e-06,
      "loss": 115.1031,
      "step": 29610
    },
    {
      "epoch": 0.24502626463167473,
      "grad_norm": 659.7858276367188,
      "learning_rate": 9.420023737338065e-06,
      "loss": 120.2869,
      "step": 29620
    },
    {
      "epoch": 0.24510898788104396,
      "grad_norm": 834.023193359375,
      "learning_rate": 9.419364246039745e-06,
      "loss": 125.4224,
      "step": 29630
    },
    {
      "epoch": 0.24519171113041321,
      "grad_norm": 1009.0372924804688,
      "learning_rate": 9.418704403112894e-06,
      "loss": 109.2442,
      "step": 29640
    },
    {
      "epoch": 0.24527443437978244,
      "grad_norm": 2674.98193359375,
      "learning_rate": 9.418044208610013e-06,
      "loss": 156.5225,
      "step": 29650
    },
    {
      "epoch": 0.24535715762915167,
      "grad_norm": 1011.0217895507812,
      "learning_rate": 9.41738366258363e-06,
      "loss": 126.1811,
      "step": 29660
    },
    {
      "epoch": 0.24543988087852092,
      "grad_norm": 610.7017211914062,
      "learning_rate": 9.416722765086304e-06,
      "loss": 144.7449,
      "step": 29670
    },
    {
      "epoch": 0.24552260412789015,
      "grad_norm": 1031.1539306640625,
      "learning_rate": 9.416061516170615e-06,
      "loss": 108.1692,
      "step": 29680
    },
    {
      "epoch": 0.24560532737725937,
      "grad_norm": 1801.7032470703125,
      "learning_rate": 9.415399915889179e-06,
      "loss": 121.3443,
      "step": 29690
    },
    {
      "epoch": 0.2456880506266286,
      "grad_norm": 1428.9144287109375,
      "learning_rate": 9.414737964294636e-06,
      "loss": 116.9526,
      "step": 29700
    },
    {
      "epoch": 0.24577077387599786,
      "grad_norm": 937.070556640625,
      "learning_rate": 9.414075661439653e-06,
      "loss": 111.9231,
      "step": 29710
    },
    {
      "epoch": 0.24585349712536708,
      "grad_norm": 1381.5968017578125,
      "learning_rate": 9.413413007376928e-06,
      "loss": 163.5947,
      "step": 29720
    },
    {
      "epoch": 0.2459362203747363,
      "grad_norm": 1236.41552734375,
      "learning_rate": 9.412750002159186e-06,
      "loss": 110.7294,
      "step": 29730
    },
    {
      "epoch": 0.24601894362410556,
      "grad_norm": 757.4229125976562,
      "learning_rate": 9.412086645839177e-06,
      "loss": 88.9742,
      "step": 29740
    },
    {
      "epoch": 0.2461016668734748,
      "grad_norm": 1002.5419311523438,
      "learning_rate": 9.411422938469683e-06,
      "loss": 137.723,
      "step": 29750
    },
    {
      "epoch": 0.24618439012284402,
      "grad_norm": 636.2584228515625,
      "learning_rate": 9.41075888010351e-06,
      "loss": 108.3714,
      "step": 29760
    },
    {
      "epoch": 0.24626711337221327,
      "grad_norm": 737.2359008789062,
      "learning_rate": 9.410094470793497e-06,
      "loss": 135.6444,
      "step": 29770
    },
    {
      "epoch": 0.2463498366215825,
      "grad_norm": 1887.1973876953125,
      "learning_rate": 9.409429710592505e-06,
      "loss": 126.0426,
      "step": 29780
    },
    {
      "epoch": 0.24643255987095172,
      "grad_norm": 1311.354248046875,
      "learning_rate": 9.408764599553429e-06,
      "loss": 156.2838,
      "step": 29790
    },
    {
      "epoch": 0.24651528312032098,
      "grad_norm": 1149.238525390625,
      "learning_rate": 9.408099137729188e-06,
      "loss": 130.7976,
      "step": 29800
    },
    {
      "epoch": 0.2465980063696902,
      "grad_norm": 1126.7828369140625,
      "learning_rate": 9.407433325172727e-06,
      "loss": 153.4184,
      "step": 29810
    },
    {
      "epoch": 0.24668072961905943,
      "grad_norm": 1250.0152587890625,
      "learning_rate": 9.406767161937025e-06,
      "loss": 142.3581,
      "step": 29820
    },
    {
      "epoch": 0.24676345286842868,
      "grad_norm": 1395.9898681640625,
      "learning_rate": 9.406100648075084e-06,
      "loss": 122.4098,
      "step": 29830
    },
    {
      "epoch": 0.2468461761177979,
      "grad_norm": 1577.0543212890625,
      "learning_rate": 9.405433783639936e-06,
      "loss": 112.6034,
      "step": 29840
    },
    {
      "epoch": 0.24692889936716714,
      "grad_norm": 1184.2935791015625,
      "learning_rate": 9.40476656868464e-06,
      "loss": 148.0747,
      "step": 29850
    },
    {
      "epoch": 0.24701162261653636,
      "grad_norm": 805.5813598632812,
      "learning_rate": 9.404099003262282e-06,
      "loss": 155.5525,
      "step": 29860
    },
    {
      "epoch": 0.24709434586590562,
      "grad_norm": 1295.2099609375,
      "learning_rate": 9.40343108742598e-06,
      "loss": 149.4768,
      "step": 29870
    },
    {
      "epoch": 0.24717706911527484,
      "grad_norm": 995.8720092773438,
      "learning_rate": 9.402762821228875e-06,
      "loss": 140.2816,
      "step": 29880
    },
    {
      "epoch": 0.24725979236464407,
      "grad_norm": 866.3977661132812,
      "learning_rate": 9.402094204724138e-06,
      "loss": 129.4959,
      "step": 29890
    },
    {
      "epoch": 0.24734251561401333,
      "grad_norm": 1852.4481201171875,
      "learning_rate": 9.401425237964966e-06,
      "loss": 102.9619,
      "step": 29900
    },
    {
      "epoch": 0.24742523886338255,
      "grad_norm": 743.556640625,
      "learning_rate": 9.400755921004592e-06,
      "loss": 85.2109,
      "step": 29910
    },
    {
      "epoch": 0.24750796211275178,
      "grad_norm": 809.3905639648438,
      "learning_rate": 9.400086253896264e-06,
      "loss": 106.4736,
      "step": 29920
    },
    {
      "epoch": 0.24759068536212103,
      "grad_norm": 783.1993408203125,
      "learning_rate": 9.399416236693264e-06,
      "loss": 125.8943,
      "step": 29930
    },
    {
      "epoch": 0.24767340861149026,
      "grad_norm": 772.568115234375,
      "learning_rate": 9.398745869448909e-06,
      "loss": 123.8559,
      "step": 29940
    },
    {
      "epoch": 0.24775613186085949,
      "grad_norm": 1002.6859130859375,
      "learning_rate": 9.39807515221653e-06,
      "loss": 101.8871,
      "step": 29950
    },
    {
      "epoch": 0.24783885511022874,
      "grad_norm": 1258.36572265625,
      "learning_rate": 9.397404085049496e-06,
      "loss": 98.4138,
      "step": 29960
    },
    {
      "epoch": 0.24792157835959797,
      "grad_norm": 1145.0703125,
      "learning_rate": 9.3967326680012e-06,
      "loss": 118.3927,
      "step": 29970
    },
    {
      "epoch": 0.2480043016089672,
      "grad_norm": 1031.3804931640625,
      "learning_rate": 9.396060901125064e-06,
      "loss": 105.3649,
      "step": 29980
    },
    {
      "epoch": 0.24808702485833645,
      "grad_norm": 2366.289794921875,
      "learning_rate": 9.395388784474538e-06,
      "loss": 168.2479,
      "step": 29990
    },
    {
      "epoch": 0.24816974810770567,
      "grad_norm": 369.7084045410156,
      "learning_rate": 9.394716318103098e-06,
      "loss": 121.3149,
      "step": 30000
    },
    {
      "epoch": 0.2482524713570749,
      "grad_norm": 1409.655029296875,
      "learning_rate": 9.394043502064249e-06,
      "loss": 105.2097,
      "step": 30010
    },
    {
      "epoch": 0.24833519460644415,
      "grad_norm": 1403.9825439453125,
      "learning_rate": 9.393370336411527e-06,
      "loss": 147.3934,
      "step": 30020
    },
    {
      "epoch": 0.24841791785581338,
      "grad_norm": 929.0065307617188,
      "learning_rate": 9.392696821198488e-06,
      "loss": 124.7842,
      "step": 30030
    },
    {
      "epoch": 0.2485006411051826,
      "grad_norm": 828.625244140625,
      "learning_rate": 9.392022956478724e-06,
      "loss": 112.9368,
      "step": 30040
    },
    {
      "epoch": 0.24858336435455183,
      "grad_norm": 3000.644287109375,
      "learning_rate": 9.391348742305849e-06,
      "loss": 148.1125,
      "step": 30050
    },
    {
      "epoch": 0.2486660876039211,
      "grad_norm": 908.979248046875,
      "learning_rate": 9.390674178733508e-06,
      "loss": 109.6535,
      "step": 30060
    },
    {
      "epoch": 0.24874881085329031,
      "grad_norm": 1394.7421875,
      "learning_rate": 9.389999265815373e-06,
      "loss": 112.9092,
      "step": 30070
    },
    {
      "epoch": 0.24883153410265954,
      "grad_norm": 1016.4574584960938,
      "learning_rate": 9.389324003605144e-06,
      "loss": 168.127,
      "step": 30080
    },
    {
      "epoch": 0.2489142573520288,
      "grad_norm": 1174.79443359375,
      "learning_rate": 9.388648392156547e-06,
      "loss": 112.0588,
      "step": 30090
    },
    {
      "epoch": 0.24899698060139802,
      "grad_norm": 2049.481689453125,
      "learning_rate": 9.387972431523341e-06,
      "loss": 127.4066,
      "step": 30100
    },
    {
      "epoch": 0.24907970385076725,
      "grad_norm": 712.5939331054688,
      "learning_rate": 9.387296121759305e-06,
      "loss": 98.8517,
      "step": 30110
    },
    {
      "epoch": 0.2491624271001365,
      "grad_norm": 872.814208984375,
      "learning_rate": 9.386619462918254e-06,
      "loss": 100.3602,
      "step": 30120
    },
    {
      "epoch": 0.24924515034950573,
      "grad_norm": 649.6997680664062,
      "learning_rate": 9.385942455054022e-06,
      "loss": 119.4873,
      "step": 30130
    },
    {
      "epoch": 0.24932787359887496,
      "grad_norm": 655.9243774414062,
      "learning_rate": 9.385265098220478e-06,
      "loss": 124.5341,
      "step": 30140
    },
    {
      "epoch": 0.2494105968482442,
      "grad_norm": 822.2200927734375,
      "learning_rate": 9.384587392471516e-06,
      "loss": 162.9077,
      "step": 30150
    },
    {
      "epoch": 0.24949332009761344,
      "grad_norm": 940.898193359375,
      "learning_rate": 9.383909337861058e-06,
      "loss": 118.583,
      "step": 30160
    },
    {
      "epoch": 0.24957604334698266,
      "grad_norm": 1128.41943359375,
      "learning_rate": 9.383230934443053e-06,
      "loss": 136.6669,
      "step": 30170
    },
    {
      "epoch": 0.24965876659635192,
      "grad_norm": 631.8690795898438,
      "learning_rate": 9.382552182271478e-06,
      "loss": 97.5566,
      "step": 30180
    },
    {
      "epoch": 0.24974148984572114,
      "grad_norm": 1021.1989135742188,
      "learning_rate": 9.38187308140034e-06,
      "loss": 146.495,
      "step": 30190
    },
    {
      "epoch": 0.24982421309509037,
      "grad_norm": 1181.3828125,
      "learning_rate": 9.381193631883672e-06,
      "loss": 150.6252,
      "step": 30200
    },
    {
      "epoch": 0.24990693634445962,
      "grad_norm": 814.0835571289062,
      "learning_rate": 9.380513833775531e-06,
      "loss": 114.7124,
      "step": 30210
    },
    {
      "epoch": 0.24998965959382885,
      "grad_norm": 1297.4193115234375,
      "learning_rate": 9.37983368713001e-06,
      "loss": 97.1973,
      "step": 30220
    },
    {
      "epoch": 0.2500723828431981,
      "grad_norm": 800.78564453125,
      "learning_rate": 9.379153192001223e-06,
      "loss": 98.411,
      "step": 30230
    },
    {
      "epoch": 0.2501551060925673,
      "grad_norm": 1123.505859375,
      "learning_rate": 9.378472348443315e-06,
      "loss": 119.3296,
      "step": 30240
    },
    {
      "epoch": 0.25023782934193656,
      "grad_norm": 888.609375,
      "learning_rate": 9.377791156510456e-06,
      "loss": 74.0182,
      "step": 30250
    },
    {
      "epoch": 0.2503205525913058,
      "grad_norm": 713.3021240234375,
      "learning_rate": 9.377109616256846e-06,
      "loss": 147.7178,
      "step": 30260
    },
    {
      "epoch": 0.250403275840675,
      "grad_norm": 1878.27880859375,
      "learning_rate": 9.37642772773671e-06,
      "loss": 154.91,
      "step": 30270
    },
    {
      "epoch": 0.25048599909004426,
      "grad_norm": 625.2662353515625,
      "learning_rate": 9.375745491004307e-06,
      "loss": 90.2972,
      "step": 30280
    },
    {
      "epoch": 0.2505687223394135,
      "grad_norm": 621.6907958984375,
      "learning_rate": 9.375062906113916e-06,
      "loss": 126.9956,
      "step": 30290
    },
    {
      "epoch": 0.2506514455887827,
      "grad_norm": 1250.3077392578125,
      "learning_rate": 9.37437997311985e-06,
      "loss": 107.0872,
      "step": 30300
    },
    {
      "epoch": 0.25073416883815197,
      "grad_norm": 720.9534912109375,
      "learning_rate": 9.373696692076446e-06,
      "loss": 105.0815,
      "step": 30310
    },
    {
      "epoch": 0.25081689208752117,
      "grad_norm": 684.220703125,
      "learning_rate": 9.373013063038066e-06,
      "loss": 129.8487,
      "step": 30320
    },
    {
      "epoch": 0.2508996153368904,
      "grad_norm": 1063.6759033203125,
      "learning_rate": 9.372329086059108e-06,
      "loss": 135.9542,
      "step": 30330
    },
    {
      "epoch": 0.2509823385862597,
      "grad_norm": 1607.3919677734375,
      "learning_rate": 9.37164476119399e-06,
      "loss": 142.6617,
      "step": 30340
    },
    {
      "epoch": 0.2510650618356289,
      "grad_norm": 542.721435546875,
      "learning_rate": 9.370960088497162e-06,
      "loss": 106.839,
      "step": 30350
    },
    {
      "epoch": 0.25114778508499813,
      "grad_norm": 549.24560546875,
      "learning_rate": 9.370275068023097e-06,
      "loss": 129.0447,
      "step": 30360
    },
    {
      "epoch": 0.2512305083343674,
      "grad_norm": 960.9791259765625,
      "learning_rate": 9.369589699826306e-06,
      "loss": 140.4398,
      "step": 30370
    },
    {
      "epoch": 0.2513132315837366,
      "grad_norm": 1057.4302978515625,
      "learning_rate": 9.368903983961315e-06,
      "loss": 138.126,
      "step": 30380
    },
    {
      "epoch": 0.25139595483310584,
      "grad_norm": 1365.5726318359375,
      "learning_rate": 9.368217920482684e-06,
      "loss": 139.174,
      "step": 30390
    },
    {
      "epoch": 0.2514786780824751,
      "grad_norm": 887.7735595703125,
      "learning_rate": 9.367531509445001e-06,
      "loss": 129.102,
      "step": 30400
    },
    {
      "epoch": 0.2515614013318443,
      "grad_norm": 1251.561767578125,
      "learning_rate": 9.366844750902878e-06,
      "loss": 121.9665,
      "step": 30410
    },
    {
      "epoch": 0.25164412458121355,
      "grad_norm": 881.740234375,
      "learning_rate": 9.36615764491096e-06,
      "loss": 82.5869,
      "step": 30420
    },
    {
      "epoch": 0.2517268478305828,
      "grad_norm": 821.780029296875,
      "learning_rate": 9.365470191523917e-06,
      "loss": 146.2663,
      "step": 30430
    },
    {
      "epoch": 0.251809571079952,
      "grad_norm": 626.3407592773438,
      "learning_rate": 9.364782390796446e-06,
      "loss": 86.4238,
      "step": 30440
    },
    {
      "epoch": 0.25189229432932125,
      "grad_norm": 1124.6002197265625,
      "learning_rate": 9.364094242783272e-06,
      "loss": 146.8187,
      "step": 30450
    },
    {
      "epoch": 0.2519750175786905,
      "grad_norm": 631.0712890625,
      "learning_rate": 9.363405747539147e-06,
      "loss": 98.5037,
      "step": 30460
    },
    {
      "epoch": 0.2520577408280597,
      "grad_norm": 949.3443603515625,
      "learning_rate": 9.362716905118851e-06,
      "loss": 139.6968,
      "step": 30470
    },
    {
      "epoch": 0.25214046407742896,
      "grad_norm": 513.1497802734375,
      "learning_rate": 9.362027715577195e-06,
      "loss": 118.3806,
      "step": 30480
    },
    {
      "epoch": 0.2522231873267982,
      "grad_norm": 1057.8067626953125,
      "learning_rate": 9.361338178969012e-06,
      "loss": 108.9348,
      "step": 30490
    },
    {
      "epoch": 0.2523059105761674,
      "grad_norm": 903.6969604492188,
      "learning_rate": 9.360648295349165e-06,
      "loss": 105.4085,
      "step": 30500
    },
    {
      "epoch": 0.25238863382553667,
      "grad_norm": 2535.44189453125,
      "learning_rate": 9.359958064772547e-06,
      "loss": 161.6714,
      "step": 30510
    },
    {
      "epoch": 0.2524713570749059,
      "grad_norm": 1677.7100830078125,
      "learning_rate": 9.359267487294075e-06,
      "loss": 128.2102,
      "step": 30520
    },
    {
      "epoch": 0.2525540803242751,
      "grad_norm": 1912.9716796875,
      "learning_rate": 9.358576562968695e-06,
      "loss": 118.4899,
      "step": 30530
    },
    {
      "epoch": 0.2526368035736444,
      "grad_norm": 763.476318359375,
      "learning_rate": 9.357885291851382e-06,
      "loss": 124.9722,
      "step": 30540
    },
    {
      "epoch": 0.25271952682301363,
      "grad_norm": 1949.4473876953125,
      "learning_rate": 9.357193673997133e-06,
      "loss": 104.1943,
      "step": 30550
    },
    {
      "epoch": 0.25280225007238283,
      "grad_norm": 1901.4537353515625,
      "learning_rate": 9.356501709460984e-06,
      "loss": 108.5047,
      "step": 30560
    },
    {
      "epoch": 0.2528849733217521,
      "grad_norm": 773.7173461914062,
      "learning_rate": 9.355809398297986e-06,
      "loss": 95.4959,
      "step": 30570
    },
    {
      "epoch": 0.25296769657112134,
      "grad_norm": 1136.826171875,
      "learning_rate": 9.355116740563225e-06,
      "loss": 136.653,
      "step": 30580
    },
    {
      "epoch": 0.25305041982049054,
      "grad_norm": 826.9710693359375,
      "learning_rate": 9.354423736311813e-06,
      "loss": 119.1377,
      "step": 30590
    },
    {
      "epoch": 0.2531331430698598,
      "grad_norm": 1087.69677734375,
      "learning_rate": 9.353730385598887e-06,
      "loss": 101.276,
      "step": 30600
    },
    {
      "epoch": 0.25321586631922904,
      "grad_norm": 567.2242431640625,
      "learning_rate": 9.353036688479615e-06,
      "loss": 116.7849,
      "step": 30610
    },
    {
      "epoch": 0.25329858956859824,
      "grad_norm": 1647.7808837890625,
      "learning_rate": 9.352342645009193e-06,
      "loss": 142.3532,
      "step": 30620
    },
    {
      "epoch": 0.2533813128179675,
      "grad_norm": 1223.5712890625,
      "learning_rate": 9.35164825524284e-06,
      "loss": 106.8768,
      "step": 30630
    },
    {
      "epoch": 0.25346403606733675,
      "grad_norm": 1215.446044921875,
      "learning_rate": 9.350953519235807e-06,
      "loss": 142.7279,
      "step": 30640
    },
    {
      "epoch": 0.25354675931670595,
      "grad_norm": 1093.0865478515625,
      "learning_rate": 9.35025843704337e-06,
      "loss": 133.1846,
      "step": 30650
    },
    {
      "epoch": 0.2536294825660752,
      "grad_norm": 603.0365600585938,
      "learning_rate": 9.349563008720836e-06,
      "loss": 143.9578,
      "step": 30660
    },
    {
      "epoch": 0.25371220581544446,
      "grad_norm": 926.9697265625,
      "learning_rate": 9.348867234323534e-06,
      "loss": 115.379,
      "step": 30670
    },
    {
      "epoch": 0.25379492906481366,
      "grad_norm": 1196.4434814453125,
      "learning_rate": 9.348171113906826e-06,
      "loss": 128.1764,
      "step": 30680
    },
    {
      "epoch": 0.2538776523141829,
      "grad_norm": 750.9150390625,
      "learning_rate": 9.347474647526095e-06,
      "loss": 194.924,
      "step": 30690
    },
    {
      "epoch": 0.2539603755635521,
      "grad_norm": 1341.235595703125,
      "learning_rate": 9.34677783523676e-06,
      "loss": 137.8295,
      "step": 30700
    },
    {
      "epoch": 0.25404309881292136,
      "grad_norm": 1126.3736572265625,
      "learning_rate": 9.346080677094262e-06,
      "loss": 132.7227,
      "step": 30710
    },
    {
      "epoch": 0.2541258220622906,
      "grad_norm": 824.2003784179688,
      "learning_rate": 9.345383173154072e-06,
      "loss": 133.4808,
      "step": 30720
    },
    {
      "epoch": 0.2542085453116598,
      "grad_norm": 929.5851440429688,
      "learning_rate": 9.344685323471682e-06,
      "loss": 109.8865,
      "step": 30730
    },
    {
      "epoch": 0.25429126856102907,
      "grad_norm": 763.9591674804688,
      "learning_rate": 9.343987128102624e-06,
      "loss": 114.478,
      "step": 30740
    },
    {
      "epoch": 0.2543739918103983,
      "grad_norm": 896.4277954101562,
      "learning_rate": 9.343288587102444e-06,
      "loss": 139.7716,
      "step": 30750
    },
    {
      "epoch": 0.2544567150597675,
      "grad_norm": 890.5599975585938,
      "learning_rate": 9.342589700526725e-06,
      "loss": 119.8424,
      "step": 30760
    },
    {
      "epoch": 0.2545394383091368,
      "grad_norm": 954.55322265625,
      "learning_rate": 9.341890468431072e-06,
      "loss": 197.9463,
      "step": 30770
    },
    {
      "epoch": 0.25462216155850603,
      "grad_norm": 962.0372314453125,
      "learning_rate": 9.341190890871123e-06,
      "loss": 173.233,
      "step": 30780
    },
    {
      "epoch": 0.25470488480787523,
      "grad_norm": 1072.42724609375,
      "learning_rate": 9.340490967902535e-06,
      "loss": 114.3112,
      "step": 30790
    },
    {
      "epoch": 0.2547876080572445,
      "grad_norm": 927.2454223632812,
      "learning_rate": 9.339790699581004e-06,
      "loss": 98.9923,
      "step": 30800
    },
    {
      "epoch": 0.25487033130661374,
      "grad_norm": 485.0035400390625,
      "learning_rate": 9.339090085962244e-06,
      "loss": 109.2545,
      "step": 30810
    },
    {
      "epoch": 0.25495305455598294,
      "grad_norm": 1289.7406005859375,
      "learning_rate": 9.338389127101998e-06,
      "loss": 137.8362,
      "step": 30820
    },
    {
      "epoch": 0.2550357778053522,
      "grad_norm": 655.8922119140625,
      "learning_rate": 9.337687823056041e-06,
      "loss": 101.9889,
      "step": 30830
    },
    {
      "epoch": 0.25511850105472145,
      "grad_norm": 1242.1337890625,
      "learning_rate": 9.336986173880169e-06,
      "loss": 106.3836,
      "step": 30840
    },
    {
      "epoch": 0.25520122430409065,
      "grad_norm": 580.3970947265625,
      "learning_rate": 9.336284179630215e-06,
      "loss": 94.8493,
      "step": 30850
    },
    {
      "epoch": 0.2552839475534599,
      "grad_norm": 789.4066162109375,
      "learning_rate": 9.335581840362026e-06,
      "loss": 74.5354,
      "step": 30860
    },
    {
      "epoch": 0.25536667080282915,
      "grad_norm": 1426.3404541015625,
      "learning_rate": 9.33487915613149e-06,
      "loss": 108.0914,
      "step": 30870
    },
    {
      "epoch": 0.25544939405219835,
      "grad_norm": 826.1908569335938,
      "learning_rate": 9.334176126994512e-06,
      "loss": 109.946,
      "step": 30880
    },
    {
      "epoch": 0.2555321173015676,
      "grad_norm": 755.2938232421875,
      "learning_rate": 9.333472753007031e-06,
      "loss": 111.354,
      "step": 30890
    },
    {
      "epoch": 0.25561484055093686,
      "grad_norm": 780.1597290039062,
      "learning_rate": 9.332769034225012e-06,
      "loss": 142.2512,
      "step": 30900
    },
    {
      "epoch": 0.25569756380030606,
      "grad_norm": 731.4649047851562,
      "learning_rate": 9.332064970704445e-06,
      "loss": 156.2841,
      "step": 30910
    },
    {
      "epoch": 0.2557802870496753,
      "grad_norm": 1020.5748291015625,
      "learning_rate": 9.33136056250135e-06,
      "loss": 127.7295,
      "step": 30920
    },
    {
      "epoch": 0.25586301029904457,
      "grad_norm": 698.0821533203125,
      "learning_rate": 9.330655809671773e-06,
      "loss": 92.2535,
      "step": 30930
    },
    {
      "epoch": 0.25594573354841377,
      "grad_norm": 698.9208374023438,
      "learning_rate": 9.32995071227179e-06,
      "loss": 124.9385,
      "step": 30940
    },
    {
      "epoch": 0.256028456797783,
      "grad_norm": 1320.0196533203125,
      "learning_rate": 9.3292452703575e-06,
      "loss": 128.1827,
      "step": 30950
    },
    {
      "epoch": 0.2561111800471523,
      "grad_norm": 2012.84033203125,
      "learning_rate": 9.328539483985031e-06,
      "loss": 155.1285,
      "step": 30960
    },
    {
      "epoch": 0.2561939032965215,
      "grad_norm": 1151.90625,
      "learning_rate": 9.327833353210541e-06,
      "loss": 111.3364,
      "step": 30970
    },
    {
      "epoch": 0.25627662654589073,
      "grad_norm": 1791.881103515625,
      "learning_rate": 9.327126878090214e-06,
      "loss": 118.88,
      "step": 30980
    },
    {
      "epoch": 0.25635934979526,
      "grad_norm": 741.9896850585938,
      "learning_rate": 9.32642005868026e-06,
      "loss": 122.3429,
      "step": 30990
    },
    {
      "epoch": 0.2564420730446292,
      "grad_norm": 712.4248657226562,
      "learning_rate": 9.325712895036916e-06,
      "loss": 125.5105,
      "step": 31000
    },
    {
      "epoch": 0.25652479629399844,
      "grad_norm": 1130.374267578125,
      "learning_rate": 9.32500538721645e-06,
      "loss": 86.682,
      "step": 31010
    },
    {
      "epoch": 0.2566075195433677,
      "grad_norm": 1284.844970703125,
      "learning_rate": 9.324297535275156e-06,
      "loss": 114.092,
      "step": 31020
    },
    {
      "epoch": 0.2566902427927369,
      "grad_norm": 911.753173828125,
      "learning_rate": 9.323589339269352e-06,
      "loss": 106.8176,
      "step": 31030
    },
    {
      "epoch": 0.25677296604210614,
      "grad_norm": 3918.962646484375,
      "learning_rate": 9.322880799255385e-06,
      "loss": 160.4931,
      "step": 31040
    },
    {
      "epoch": 0.25685568929147534,
      "grad_norm": 835.4883422851562,
      "learning_rate": 9.322171915289635e-06,
      "loss": 108.218,
      "step": 31050
    },
    {
      "epoch": 0.2569384125408446,
      "grad_norm": 745.3687133789062,
      "learning_rate": 9.321462687428499e-06,
      "loss": 103.9572,
      "step": 31060
    },
    {
      "epoch": 0.25702113579021385,
      "grad_norm": 2105.262939453125,
      "learning_rate": 9.320753115728413e-06,
      "loss": 137.9624,
      "step": 31070
    },
    {
      "epoch": 0.25710385903958305,
      "grad_norm": 1638.893798828125,
      "learning_rate": 9.320043200245829e-06,
      "loss": 76.6734,
      "step": 31080
    },
    {
      "epoch": 0.2571865822889523,
      "grad_norm": 1251.9464111328125,
      "learning_rate": 9.319332941037235e-06,
      "loss": 128.9104,
      "step": 31090
    },
    {
      "epoch": 0.25726930553832156,
      "grad_norm": 946.0052490234375,
      "learning_rate": 9.31862233815914e-06,
      "loss": 83.5885,
      "step": 31100
    },
    {
      "epoch": 0.25735202878769076,
      "grad_norm": 613.86181640625,
      "learning_rate": 9.317911391668087e-06,
      "loss": 88.8766,
      "step": 31110
    },
    {
      "epoch": 0.25743475203706,
      "grad_norm": 980.6512451171875,
      "learning_rate": 9.317200101620641e-06,
      "loss": 111.9333,
      "step": 31120
    },
    {
      "epoch": 0.25751747528642926,
      "grad_norm": 959.6453857421875,
      "learning_rate": 9.316488468073397e-06,
      "loss": 87.5497,
      "step": 31130
    },
    {
      "epoch": 0.25760019853579846,
      "grad_norm": 784.59033203125,
      "learning_rate": 9.315776491082973e-06,
      "loss": 140.7631,
      "step": 31140
    },
    {
      "epoch": 0.2576829217851677,
      "grad_norm": 1294.2275390625,
      "learning_rate": 9.315064170706023e-06,
      "loss": 114.7354,
      "step": 31150
    },
    {
      "epoch": 0.25776564503453697,
      "grad_norm": 1613.03857421875,
      "learning_rate": 9.31435150699922e-06,
      "loss": 123.3567,
      "step": 31160
    },
    {
      "epoch": 0.25784836828390617,
      "grad_norm": 1174.7305908203125,
      "learning_rate": 9.313638500019267e-06,
      "loss": 133.3073,
      "step": 31170
    },
    {
      "epoch": 0.2579310915332754,
      "grad_norm": 587.86572265625,
      "learning_rate": 9.312925149822895e-06,
      "loss": 90.9177,
      "step": 31180
    },
    {
      "epoch": 0.2580138147826447,
      "grad_norm": 934.0054931640625,
      "learning_rate": 9.312211456466862e-06,
      "loss": 127.4864,
      "step": 31190
    },
    {
      "epoch": 0.2580965380320139,
      "grad_norm": 712.6873779296875,
      "learning_rate": 9.311497420007955e-06,
      "loss": 111.8241,
      "step": 31200
    },
    {
      "epoch": 0.25817926128138313,
      "grad_norm": 1259.027587890625,
      "learning_rate": 9.310783040502987e-06,
      "loss": 120.1594,
      "step": 31210
    },
    {
      "epoch": 0.2582619845307524,
      "grad_norm": 1388.41162109375,
      "learning_rate": 9.310068318008794e-06,
      "loss": 121.605,
      "step": 31220
    },
    {
      "epoch": 0.2583447077801216,
      "grad_norm": 1037.8282470703125,
      "learning_rate": 9.309353252582246e-06,
      "loss": 138.8729,
      "step": 31230
    },
    {
      "epoch": 0.25842743102949084,
      "grad_norm": 943.52490234375,
      "learning_rate": 9.308637844280236e-06,
      "loss": 132.2363,
      "step": 31240
    },
    {
      "epoch": 0.2585101542788601,
      "grad_norm": 1297.0338134765625,
      "learning_rate": 9.307922093159688e-06,
      "loss": 113.9879,
      "step": 31250
    },
    {
      "epoch": 0.2585928775282293,
      "grad_norm": 739.4756469726562,
      "learning_rate": 9.30720599927755e-06,
      "loss": 79.2995,
      "step": 31260
    },
    {
      "epoch": 0.25867560077759855,
      "grad_norm": 1136.6614990234375,
      "learning_rate": 9.306489562690797e-06,
      "loss": 148.8123,
      "step": 31270
    },
    {
      "epoch": 0.2587583240269678,
      "grad_norm": 1102.057861328125,
      "learning_rate": 9.305772783456435e-06,
      "loss": 126.115,
      "step": 31280
    },
    {
      "epoch": 0.258841047276337,
      "grad_norm": 1000.919677734375,
      "learning_rate": 9.305055661631493e-06,
      "loss": 128.0628,
      "step": 31290
    },
    {
      "epoch": 0.25892377052570625,
      "grad_norm": 1486.086669921875,
      "learning_rate": 9.304338197273029e-06,
      "loss": 141.9742,
      "step": 31300
    },
    {
      "epoch": 0.2590064937750755,
      "grad_norm": 1208.7861328125,
      "learning_rate": 9.303620390438128e-06,
      "loss": 119.3574,
      "step": 31310
    },
    {
      "epoch": 0.2590892170244447,
      "grad_norm": 1793.4461669921875,
      "learning_rate": 9.302902241183905e-06,
      "loss": 115.7504,
      "step": 31320
    },
    {
      "epoch": 0.25917194027381396,
      "grad_norm": 1034.2620849609375,
      "learning_rate": 9.302183749567498e-06,
      "loss": 104.3807,
      "step": 31330
    },
    {
      "epoch": 0.2592546635231832,
      "grad_norm": 1191.4996337890625,
      "learning_rate": 9.301464915646074e-06,
      "loss": 95.0326,
      "step": 31340
    },
    {
      "epoch": 0.2593373867725524,
      "grad_norm": 663.1774291992188,
      "learning_rate": 9.30074573947683e-06,
      "loss": 105.2758,
      "step": 31350
    },
    {
      "epoch": 0.25942011002192167,
      "grad_norm": 1111.489501953125,
      "learning_rate": 9.30002622111698e-06,
      "loss": 120.7122,
      "step": 31360
    },
    {
      "epoch": 0.2595028332712909,
      "grad_norm": 1140.2496337890625,
      "learning_rate": 9.299306360623782e-06,
      "loss": 111.4868,
      "step": 31370
    },
    {
      "epoch": 0.2595855565206601,
      "grad_norm": 1038.2596435546875,
      "learning_rate": 9.298586158054508e-06,
      "loss": 119.8149,
      "step": 31380
    },
    {
      "epoch": 0.2596682797700294,
      "grad_norm": 1008.992431640625,
      "learning_rate": 9.297865613466459e-06,
      "loss": 145.3494,
      "step": 31390
    },
    {
      "epoch": 0.25975100301939863,
      "grad_norm": 1030.713623046875,
      "learning_rate": 9.29714472691697e-06,
      "loss": 116.2307,
      "step": 31400
    },
    {
      "epoch": 0.25983372626876783,
      "grad_norm": 1515.673095703125,
      "learning_rate": 9.296423498463396e-06,
      "loss": 125.331,
      "step": 31410
    },
    {
      "epoch": 0.2599164495181371,
      "grad_norm": 1273.73486328125,
      "learning_rate": 9.29570192816312e-06,
      "loss": 140.6214,
      "step": 31420
    },
    {
      "epoch": 0.2599991727675063,
      "grad_norm": 710.17236328125,
      "learning_rate": 9.29498001607356e-06,
      "loss": 108.083,
      "step": 31430
    },
    {
      "epoch": 0.26008189601687554,
      "grad_norm": 900.10107421875,
      "learning_rate": 9.294257762252148e-06,
      "loss": 98.9134,
      "step": 31440
    },
    {
      "epoch": 0.2601646192662448,
      "grad_norm": 875.4248657226562,
      "learning_rate": 9.293535166756356e-06,
      "loss": 174.0914,
      "step": 31450
    },
    {
      "epoch": 0.260247342515614,
      "grad_norm": 1023.4577026367188,
      "learning_rate": 9.292812229643674e-06,
      "loss": 96.2018,
      "step": 31460
    },
    {
      "epoch": 0.26033006576498324,
      "grad_norm": 1871.7161865234375,
      "learning_rate": 9.292088950971624e-06,
      "loss": 135.2347,
      "step": 31470
    },
    {
      "epoch": 0.2604127890143525,
      "grad_norm": 1158.376953125,
      "learning_rate": 9.291365330797755e-06,
      "loss": 131.9809,
      "step": 31480
    },
    {
      "epoch": 0.2604955122637217,
      "grad_norm": 962.1968383789062,
      "learning_rate": 9.290641369179643e-06,
      "loss": 109.7965,
      "step": 31490
    },
    {
      "epoch": 0.26057823551309095,
      "grad_norm": 4627.09521484375,
      "learning_rate": 9.289917066174887e-06,
      "loss": 133.032,
      "step": 31500
    },
    {
      "epoch": 0.2606609587624602,
      "grad_norm": 1069.57177734375,
      "learning_rate": 9.289192421841116e-06,
      "loss": 114.866,
      "step": 31510
    },
    {
      "epoch": 0.2607436820118294,
      "grad_norm": 1533.0517578125,
      "learning_rate": 9.288467436235992e-06,
      "loss": 135.5069,
      "step": 31520
    },
    {
      "epoch": 0.26082640526119866,
      "grad_norm": 876.2843627929688,
      "learning_rate": 9.287742109417194e-06,
      "loss": 165.9743,
      "step": 31530
    },
    {
      "epoch": 0.2609091285105679,
      "grad_norm": 849.820556640625,
      "learning_rate": 9.287016441442435e-06,
      "loss": 113.8865,
      "step": 31540
    },
    {
      "epoch": 0.2609918517599371,
      "grad_norm": 1214.044189453125,
      "learning_rate": 9.28629043236945e-06,
      "loss": 129.5201,
      "step": 31550
    },
    {
      "epoch": 0.26107457500930636,
      "grad_norm": 974.0680541992188,
      "learning_rate": 9.285564082256011e-06,
      "loss": 106.2931,
      "step": 31560
    },
    {
      "epoch": 0.2611572982586756,
      "grad_norm": 591.6702270507812,
      "learning_rate": 9.284837391159904e-06,
      "loss": 77.5611,
      "step": 31570
    },
    {
      "epoch": 0.2612400215080448,
      "grad_norm": 995.6359252929688,
      "learning_rate": 9.284110359138951e-06,
      "loss": 169.7267,
      "step": 31580
    },
    {
      "epoch": 0.26132274475741407,
      "grad_norm": 993.3253784179688,
      "learning_rate": 9.283382986250997e-06,
      "loss": 117.8098,
      "step": 31590
    },
    {
      "epoch": 0.2614054680067833,
      "grad_norm": 720.0477905273438,
      "learning_rate": 9.282655272553917e-06,
      "loss": 164.0745,
      "step": 31600
    },
    {
      "epoch": 0.2614881912561525,
      "grad_norm": 1000.5869750976562,
      "learning_rate": 9.281927218105613e-06,
      "loss": 103.8817,
      "step": 31610
    },
    {
      "epoch": 0.2615709145055218,
      "grad_norm": 701.7263793945312,
      "learning_rate": 9.281198822964011e-06,
      "loss": 115.0276,
      "step": 31620
    },
    {
      "epoch": 0.26165363775489103,
      "grad_norm": 1523.617919921875,
      "learning_rate": 9.280470087187066e-06,
      "loss": 150.8629,
      "step": 31630
    },
    {
      "epoch": 0.26173636100426023,
      "grad_norm": 1352.073486328125,
      "learning_rate": 9.279741010832761e-06,
      "loss": 111.8819,
      "step": 31640
    },
    {
      "epoch": 0.2618190842536295,
      "grad_norm": 784.3294677734375,
      "learning_rate": 9.279011593959107e-06,
      "loss": 134.8354,
      "step": 31650
    },
    {
      "epoch": 0.26190180750299874,
      "grad_norm": 518.8139038085938,
      "learning_rate": 9.278281836624137e-06,
      "loss": 109.1452,
      "step": 31660
    },
    {
      "epoch": 0.26198453075236794,
      "grad_norm": 738.8187255859375,
      "learning_rate": 9.277551738885915e-06,
      "loss": 137.3162,
      "step": 31670
    },
    {
      "epoch": 0.2620672540017372,
      "grad_norm": 1180.13037109375,
      "learning_rate": 9.276821300802535e-06,
      "loss": 101.0768,
      "step": 31680
    },
    {
      "epoch": 0.26214997725110645,
      "grad_norm": 1431.0516357421875,
      "learning_rate": 9.276090522432109e-06,
      "loss": 99.368,
      "step": 31690
    },
    {
      "epoch": 0.26223270050047565,
      "grad_norm": 688.6452026367188,
      "learning_rate": 9.275359403832787e-06,
      "loss": 123.8677,
      "step": 31700
    },
    {
      "epoch": 0.2623154237498449,
      "grad_norm": 854.5988159179688,
      "learning_rate": 9.274627945062738e-06,
      "loss": 111.6149,
      "step": 31710
    },
    {
      "epoch": 0.26239814699921415,
      "grad_norm": 1075.212158203125,
      "learning_rate": 9.27389614618016e-06,
      "loss": 138.4687,
      "step": 31720
    },
    {
      "epoch": 0.26248087024858335,
      "grad_norm": 573.2298583984375,
      "learning_rate": 9.273164007243281e-06,
      "loss": 108.3331,
      "step": 31730
    },
    {
      "epoch": 0.2625635934979526,
      "grad_norm": 990.3570556640625,
      "learning_rate": 9.272431528310354e-06,
      "loss": 112.5479,
      "step": 31740
    },
    {
      "epoch": 0.26264631674732186,
      "grad_norm": 1721.805419921875,
      "learning_rate": 9.271698709439658e-06,
      "loss": 140.3488,
      "step": 31750
    },
    {
      "epoch": 0.26272903999669106,
      "grad_norm": 1286.5728759765625,
      "learning_rate": 9.2709655506895e-06,
      "loss": 168.1867,
      "step": 31760
    },
    {
      "epoch": 0.2628117632460603,
      "grad_norm": 582.025146484375,
      "learning_rate": 9.270232052118214e-06,
      "loss": 119.0196,
      "step": 31770
    },
    {
      "epoch": 0.2628944864954295,
      "grad_norm": 1799.970703125,
      "learning_rate": 9.26949821378416e-06,
      "loss": 130.6446,
      "step": 31780
    },
    {
      "epoch": 0.26297720974479877,
      "grad_norm": 810.4905395507812,
      "learning_rate": 9.268764035745727e-06,
      "loss": 123.3437,
      "step": 31790
    },
    {
      "epoch": 0.263059932994168,
      "grad_norm": 1139.095947265625,
      "learning_rate": 9.268029518061335e-06,
      "loss": 138.0163,
      "step": 31800
    },
    {
      "epoch": 0.2631426562435372,
      "grad_norm": 728.7420654296875,
      "learning_rate": 9.267294660789417e-06,
      "loss": 118.9001,
      "step": 31810
    },
    {
      "epoch": 0.2632253794929065,
      "grad_norm": 937.4639282226562,
      "learning_rate": 9.26655946398845e-06,
      "loss": 101.5111,
      "step": 31820
    },
    {
      "epoch": 0.26330810274227573,
      "grad_norm": 951.8093872070312,
      "learning_rate": 9.265823927716927e-06,
      "loss": 114.6193,
      "step": 31830
    },
    {
      "epoch": 0.26339082599164493,
      "grad_norm": 1131.0379638671875,
      "learning_rate": 9.26508805203337e-06,
      "loss": 164.4967,
      "step": 31840
    },
    {
      "epoch": 0.2634735492410142,
      "grad_norm": 916.825439453125,
      "learning_rate": 9.264351836996332e-06,
      "loss": 99.9893,
      "step": 31850
    },
    {
      "epoch": 0.26355627249038344,
      "grad_norm": 781.8618774414062,
      "learning_rate": 9.26361528266439e-06,
      "loss": 147.4806,
      "step": 31860
    },
    {
      "epoch": 0.26363899573975264,
      "grad_norm": 1081.302001953125,
      "learning_rate": 9.262878389096147e-06,
      "loss": 107.9612,
      "step": 31870
    },
    {
      "epoch": 0.2637217189891219,
      "grad_norm": 1674.4130859375,
      "learning_rate": 9.262141156350233e-06,
      "loss": 120.8496,
      "step": 31880
    },
    {
      "epoch": 0.26380444223849114,
      "grad_norm": 599.2373046875,
      "learning_rate": 9.261403584485308e-06,
      "loss": 130.4039,
      "step": 31890
    },
    {
      "epoch": 0.26388716548786034,
      "grad_norm": 931.727783203125,
      "learning_rate": 9.260665673560058e-06,
      "loss": 94.2291,
      "step": 31900
    },
    {
      "epoch": 0.2639698887372296,
      "grad_norm": 695.3705444335938,
      "learning_rate": 9.259927423633193e-06,
      "loss": 173.953,
      "step": 31910
    },
    {
      "epoch": 0.26405261198659885,
      "grad_norm": 643.5379638671875,
      "learning_rate": 9.259188834763455e-06,
      "loss": 91.6798,
      "step": 31920
    },
    {
      "epoch": 0.26413533523596805,
      "grad_norm": 1540.7181396484375,
      "learning_rate": 9.258449907009607e-06,
      "loss": 126.6724,
      "step": 31930
    },
    {
      "epoch": 0.2642180584853373,
      "grad_norm": 1498.8092041015625,
      "learning_rate": 9.257710640430444e-06,
      "loss": 110.8607,
      "step": 31940
    },
    {
      "epoch": 0.26430078173470656,
      "grad_norm": 626.6985473632812,
      "learning_rate": 9.256971035084786e-06,
      "loss": 85.7513,
      "step": 31950
    },
    {
      "epoch": 0.26438350498407576,
      "grad_norm": 1363.011962890625,
      "learning_rate": 9.256231091031477e-06,
      "loss": 101.1794,
      "step": 31960
    },
    {
      "epoch": 0.264466228233445,
      "grad_norm": 1134.742919921875,
      "learning_rate": 9.255490808329397e-06,
      "loss": 212.7933,
      "step": 31970
    },
    {
      "epoch": 0.26454895148281427,
      "grad_norm": 1176.7347412109375,
      "learning_rate": 9.254750187037443e-06,
      "loss": 122.8415,
      "step": 31980
    },
    {
      "epoch": 0.26463167473218346,
      "grad_norm": 843.20458984375,
      "learning_rate": 9.254009227214543e-06,
      "loss": 115.774,
      "step": 31990
    },
    {
      "epoch": 0.2647143979815527,
      "grad_norm": 990.8587646484375,
      "learning_rate": 9.253267928919652e-06,
      "loss": 141.8495,
      "step": 32000
    },
    {
      "epoch": 0.264797121230922,
      "grad_norm": 1063.32763671875,
      "learning_rate": 9.25252629221175e-06,
      "loss": 118.555,
      "step": 32010
    },
    {
      "epoch": 0.26487984448029117,
      "grad_norm": 1893.1072998046875,
      "learning_rate": 9.251784317149848e-06,
      "loss": 122.1342,
      "step": 32020
    },
    {
      "epoch": 0.2649625677296604,
      "grad_norm": 1173.83349609375,
      "learning_rate": 9.251042003792983e-06,
      "loss": 161.0942,
      "step": 32030
    },
    {
      "epoch": 0.2650452909790297,
      "grad_norm": 927.9155883789062,
      "learning_rate": 9.250299352200214e-06,
      "loss": 78.0564,
      "step": 32040
    },
    {
      "epoch": 0.2651280142283989,
      "grad_norm": 1478.896484375,
      "learning_rate": 9.249556362430631e-06,
      "loss": 129.6906,
      "step": 32050
    },
    {
      "epoch": 0.26521073747776813,
      "grad_norm": 772.536865234375,
      "learning_rate": 9.248813034543353e-06,
      "loss": 102.5596,
      "step": 32060
    },
    {
      "epoch": 0.2652934607271374,
      "grad_norm": 728.9833984375,
      "learning_rate": 9.24806936859752e-06,
      "loss": 111.6626,
      "step": 32070
    },
    {
      "epoch": 0.2653761839765066,
      "grad_norm": 1042.043701171875,
      "learning_rate": 9.247325364652304e-06,
      "loss": 132.3886,
      "step": 32080
    },
    {
      "epoch": 0.26545890722587584,
      "grad_norm": 490.30419921875,
      "learning_rate": 9.2465810227669e-06,
      "loss": 128.4408,
      "step": 32090
    },
    {
      "epoch": 0.2655416304752451,
      "grad_norm": 1035.492919921875,
      "learning_rate": 9.245836343000534e-06,
      "loss": 102.5217,
      "step": 32100
    },
    {
      "epoch": 0.2656243537246143,
      "grad_norm": 1005.1189575195312,
      "learning_rate": 9.245091325412456e-06,
      "loss": 112.0046,
      "step": 32110
    },
    {
      "epoch": 0.26570707697398355,
      "grad_norm": 864.4721069335938,
      "learning_rate": 9.244345970061944e-06,
      "loss": 109.3595,
      "step": 32120
    },
    {
      "epoch": 0.26578980022335275,
      "grad_norm": 803.18896484375,
      "learning_rate": 9.243600277008301e-06,
      "loss": 123.4932,
      "step": 32130
    },
    {
      "epoch": 0.265872523472722,
      "grad_norm": 1339.6492919921875,
      "learning_rate": 9.24285424631086e-06,
      "loss": 131.0302,
      "step": 32140
    },
    {
      "epoch": 0.26595524672209125,
      "grad_norm": 331.6441650390625,
      "learning_rate": 9.242107878028978e-06,
      "loss": 84.0776,
      "step": 32150
    },
    {
      "epoch": 0.26603796997146045,
      "grad_norm": 1174.8720703125,
      "learning_rate": 9.241361172222043e-06,
      "loss": 144.0584,
      "step": 32160
    },
    {
      "epoch": 0.2661206932208297,
      "grad_norm": 955.9444580078125,
      "learning_rate": 9.240614128949463e-06,
      "loss": 79.2642,
      "step": 32170
    },
    {
      "epoch": 0.26620341647019896,
      "grad_norm": 408.5090637207031,
      "learning_rate": 9.239866748270679e-06,
      "loss": 80.5909,
      "step": 32180
    },
    {
      "epoch": 0.26628613971956816,
      "grad_norm": 1054.4498291015625,
      "learning_rate": 9.239119030245156e-06,
      "loss": 105.1165,
      "step": 32190
    },
    {
      "epoch": 0.2663688629689374,
      "grad_norm": 1652.845703125,
      "learning_rate": 9.238370974932387e-06,
      "loss": 144.1475,
      "step": 32200
    },
    {
      "epoch": 0.26645158621830667,
      "grad_norm": 897.6002807617188,
      "learning_rate": 9.23762258239189e-06,
      "loss": 85.8102,
      "step": 32210
    },
    {
      "epoch": 0.26653430946767587,
      "grad_norm": 848.9139404296875,
      "learning_rate": 9.236873852683213e-06,
      "loss": 107.4219,
      "step": 32220
    },
    {
      "epoch": 0.2666170327170451,
      "grad_norm": 900.8207397460938,
      "learning_rate": 9.23612478586593e-06,
      "loss": 77.9687,
      "step": 32230
    },
    {
      "epoch": 0.2666997559664144,
      "grad_norm": 1326.0458984375,
      "learning_rate": 9.235375381999636e-06,
      "loss": 106.4991,
      "step": 32240
    },
    {
      "epoch": 0.2667824792157836,
      "grad_norm": 2422.508056640625,
      "learning_rate": 9.234625641143962e-06,
      "loss": 138.341,
      "step": 32250
    },
    {
      "epoch": 0.26686520246515283,
      "grad_norm": 964.7281494140625,
      "learning_rate": 9.233875563358559e-06,
      "loss": 122.4212,
      "step": 32260
    },
    {
      "epoch": 0.2669479257145221,
      "grad_norm": 1445.0108642578125,
      "learning_rate": 9.23312514870311e-06,
      "loss": 107.632,
      "step": 32270
    },
    {
      "epoch": 0.2670306489638913,
      "grad_norm": 628.1731567382812,
      "learning_rate": 9.232374397237318e-06,
      "loss": 118.4748,
      "step": 32280
    },
    {
      "epoch": 0.26711337221326054,
      "grad_norm": 1020.5704345703125,
      "learning_rate": 9.231623309020922e-06,
      "loss": 132.8099,
      "step": 32290
    },
    {
      "epoch": 0.2671960954626298,
      "grad_norm": 949.1611938476562,
      "learning_rate": 9.230871884113679e-06,
      "loss": 128.9596,
      "step": 32300
    },
    {
      "epoch": 0.267278818711999,
      "grad_norm": 853.702880859375,
      "learning_rate": 9.230120122575376e-06,
      "loss": 117.7804,
      "step": 32310
    },
    {
      "epoch": 0.26736154196136824,
      "grad_norm": 856.49462890625,
      "learning_rate": 9.22936802446583e-06,
      "loss": 179.9062,
      "step": 32320
    },
    {
      "epoch": 0.2674442652107375,
      "grad_norm": 847.5012817382812,
      "learning_rate": 9.228615589844879e-06,
      "loss": 84.1749,
      "step": 32330
    },
    {
      "epoch": 0.2675269884601067,
      "grad_norm": 1034.2694091796875,
      "learning_rate": 9.227862818772392e-06,
      "loss": 131.7186,
      "step": 32340
    },
    {
      "epoch": 0.26760971170947595,
      "grad_norm": 1057.1470947265625,
      "learning_rate": 9.227109711308265e-06,
      "loss": 94.973,
      "step": 32350
    },
    {
      "epoch": 0.2676924349588452,
      "grad_norm": 374.79473876953125,
      "learning_rate": 9.226356267512417e-06,
      "loss": 107.7693,
      "step": 32360
    },
    {
      "epoch": 0.2677751582082144,
      "grad_norm": 817.5911865234375,
      "learning_rate": 9.225602487444799e-06,
      "loss": 107.2883,
      "step": 32370
    },
    {
      "epoch": 0.26785788145758366,
      "grad_norm": 1157.15234375,
      "learning_rate": 9.224848371165382e-06,
      "loss": 170.9429,
      "step": 32380
    },
    {
      "epoch": 0.2679406047069529,
      "grad_norm": 1658.9010009765625,
      "learning_rate": 9.224093918734172e-06,
      "loss": 202.5666,
      "step": 32390
    },
    {
      "epoch": 0.2680233279563221,
      "grad_norm": 1403.2574462890625,
      "learning_rate": 9.223339130211194e-06,
      "loss": 113.1494,
      "step": 32400
    },
    {
      "epoch": 0.26810605120569136,
      "grad_norm": 939.0480346679688,
      "learning_rate": 9.222584005656501e-06,
      "loss": 114.9759,
      "step": 32410
    },
    {
      "epoch": 0.2681887744550606,
      "grad_norm": 578.2918701171875,
      "learning_rate": 9.22182854513018e-06,
      "loss": 94.9374,
      "step": 32420
    },
    {
      "epoch": 0.2682714977044298,
      "grad_norm": 2604.21484375,
      "learning_rate": 9.221072748692336e-06,
      "loss": 109.7514,
      "step": 32430
    },
    {
      "epoch": 0.26835422095379907,
      "grad_norm": 962.8878173828125,
      "learning_rate": 9.220316616403109e-06,
      "loss": 104.4484,
      "step": 32440
    },
    {
      "epoch": 0.2684369442031683,
      "grad_norm": 759.3588256835938,
      "learning_rate": 9.219560148322655e-06,
      "loss": 101.3538,
      "step": 32450
    },
    {
      "epoch": 0.2685196674525375,
      "grad_norm": 1004.7863159179688,
      "learning_rate": 9.218803344511165e-06,
      "loss": 134.46,
      "step": 32460
    },
    {
      "epoch": 0.2686023907019068,
      "grad_norm": 1546.0279541015625,
      "learning_rate": 9.218046205028854e-06,
      "loss": 102.795,
      "step": 32470
    },
    {
      "epoch": 0.26868511395127603,
      "grad_norm": 422.1183776855469,
      "learning_rate": 9.217288729935966e-06,
      "loss": 100.5324,
      "step": 32480
    },
    {
      "epoch": 0.26876783720064523,
      "grad_norm": 953.4854736328125,
      "learning_rate": 9.216530919292768e-06,
      "loss": 147.8428,
      "step": 32490
    },
    {
      "epoch": 0.2688505604500145,
      "grad_norm": 910.6680297851562,
      "learning_rate": 9.215772773159556e-06,
      "loss": 138.0076,
      "step": 32500
    },
    {
      "epoch": 0.2689332836993837,
      "grad_norm": 1041.43505859375,
      "learning_rate": 9.215014291596653e-06,
      "loss": 120.8348,
      "step": 32510
    },
    {
      "epoch": 0.26901600694875294,
      "grad_norm": 1151.9285888671875,
      "learning_rate": 9.214255474664405e-06,
      "loss": 150.9121,
      "step": 32520
    },
    {
      "epoch": 0.2690987301981222,
      "grad_norm": 921.2622680664062,
      "learning_rate": 9.213496322423193e-06,
      "loss": 83.8476,
      "step": 32530
    },
    {
      "epoch": 0.2691814534474914,
      "grad_norm": 411.46826171875,
      "learning_rate": 9.212736834933413e-06,
      "loss": 129.1243,
      "step": 32540
    },
    {
      "epoch": 0.26926417669686065,
      "grad_norm": 3256.450927734375,
      "learning_rate": 9.211977012255497e-06,
      "loss": 87.7072,
      "step": 32550
    },
    {
      "epoch": 0.2693468999462299,
      "grad_norm": 1524.965087890625,
      "learning_rate": 9.211216854449903e-06,
      "loss": 128.8632,
      "step": 32560
    },
    {
      "epoch": 0.2694296231955991,
      "grad_norm": 953.8908081054688,
      "learning_rate": 9.210456361577109e-06,
      "loss": 113.7588,
      "step": 32570
    },
    {
      "epoch": 0.26951234644496835,
      "grad_norm": 1008.9974975585938,
      "learning_rate": 9.209695533697624e-06,
      "loss": 122.1501,
      "step": 32580
    },
    {
      "epoch": 0.2695950696943376,
      "grad_norm": 805.004150390625,
      "learning_rate": 9.208934370871989e-06,
      "loss": 113.7434,
      "step": 32590
    },
    {
      "epoch": 0.2696777929437068,
      "grad_norm": 636.7761840820312,
      "learning_rate": 9.20817287316076e-06,
      "loss": 113.1953,
      "step": 32600
    },
    {
      "epoch": 0.26976051619307606,
      "grad_norm": 763.083984375,
      "learning_rate": 9.20741104062453e-06,
      "loss": 101.9306,
      "step": 32610
    },
    {
      "epoch": 0.2698432394424453,
      "grad_norm": 1067.39208984375,
      "learning_rate": 9.206648873323912e-06,
      "loss": 104.6595,
      "step": 32620
    },
    {
      "epoch": 0.2699259626918145,
      "grad_norm": 1282.15576171875,
      "learning_rate": 9.205886371319548e-06,
      "loss": 112.2921,
      "step": 32630
    },
    {
      "epoch": 0.27000868594118377,
      "grad_norm": 1269.3165283203125,
      "learning_rate": 9.20512353467211e-06,
      "loss": 101.4175,
      "step": 32640
    },
    {
      "epoch": 0.270091409190553,
      "grad_norm": 1416.7222900390625,
      "learning_rate": 9.204360363442288e-06,
      "loss": 94.3014,
      "step": 32650
    },
    {
      "epoch": 0.2701741324399222,
      "grad_norm": 577.2730712890625,
      "learning_rate": 9.20359685769081e-06,
      "loss": 178.5912,
      "step": 32660
    },
    {
      "epoch": 0.2702568556892915,
      "grad_norm": 976.02880859375,
      "learning_rate": 9.202833017478421e-06,
      "loss": 142.0586,
      "step": 32670
    },
    {
      "epoch": 0.27033957893866073,
      "grad_norm": 1126.535888671875,
      "learning_rate": 9.2020688428659e-06,
      "loss": 100.6238,
      "step": 32680
    },
    {
      "epoch": 0.27042230218802993,
      "grad_norm": 945.1123657226562,
      "learning_rate": 9.201304333914042e-06,
      "loss": 124.4394,
      "step": 32690
    },
    {
      "epoch": 0.2705050254373992,
      "grad_norm": 600.9089965820312,
      "learning_rate": 9.200539490683682e-06,
      "loss": 93.1028,
      "step": 32700
    },
    {
      "epoch": 0.27058774868676844,
      "grad_norm": 704.4324951171875,
      "learning_rate": 9.19977431323567e-06,
      "loss": 124.6708,
      "step": 32710
    },
    {
      "epoch": 0.27067047193613764,
      "grad_norm": 938.8729858398438,
      "learning_rate": 9.199008801630893e-06,
      "loss": 119.1146,
      "step": 32720
    },
    {
      "epoch": 0.2707531951855069,
      "grad_norm": 1016.2319946289062,
      "learning_rate": 9.198242955930257e-06,
      "loss": 126.7218,
      "step": 32730
    },
    {
      "epoch": 0.27083591843487614,
      "grad_norm": 832.0853881835938,
      "learning_rate": 9.197476776194693e-06,
      "loss": 102.6724,
      "step": 32740
    },
    {
      "epoch": 0.27091864168424534,
      "grad_norm": 1260.37548828125,
      "learning_rate": 9.196710262485168e-06,
      "loss": 107.5099,
      "step": 32750
    },
    {
      "epoch": 0.2710013649336146,
      "grad_norm": 840.14990234375,
      "learning_rate": 9.195943414862667e-06,
      "loss": 124.9764,
      "step": 32760
    },
    {
      "epoch": 0.27108408818298385,
      "grad_norm": 486.6063537597656,
      "learning_rate": 9.195176233388206e-06,
      "loss": 92.0499,
      "step": 32770
    },
    {
      "epoch": 0.27116681143235305,
      "grad_norm": 907.8724975585938,
      "learning_rate": 9.194408718122825e-06,
      "loss": 120.6719,
      "step": 32780
    },
    {
      "epoch": 0.2712495346817223,
      "grad_norm": 572.4683837890625,
      "learning_rate": 9.193640869127592e-06,
      "loss": 124.2721,
      "step": 32790
    },
    {
      "epoch": 0.27133225793109156,
      "grad_norm": 957.0466918945312,
      "learning_rate": 9.192872686463601e-06,
      "loss": 131.9941,
      "step": 32800
    },
    {
      "epoch": 0.27141498118046076,
      "grad_norm": 691.4140014648438,
      "learning_rate": 9.192104170191973e-06,
      "loss": 101.0082,
      "step": 32810
    },
    {
      "epoch": 0.27149770442983,
      "grad_norm": 791.294677734375,
      "learning_rate": 9.191335320373856e-06,
      "loss": 138.9451,
      "step": 32820
    },
    {
      "epoch": 0.27158042767919927,
      "grad_norm": 1051.594482421875,
      "learning_rate": 9.190566137070422e-06,
      "loss": 107.2597,
      "step": 32830
    },
    {
      "epoch": 0.27166315092856846,
      "grad_norm": 1549.462158203125,
      "learning_rate": 9.189796620342875e-06,
      "loss": 153.74,
      "step": 32840
    },
    {
      "epoch": 0.2717458741779377,
      "grad_norm": 751.6979370117188,
      "learning_rate": 9.189026770252437e-06,
      "loss": 108.3263,
      "step": 32850
    },
    {
      "epoch": 0.2718285974273069,
      "grad_norm": 4171.58203125,
      "learning_rate": 9.188256586860365e-06,
      "loss": 177.8506,
      "step": 32860
    },
    {
      "epoch": 0.27191132067667617,
      "grad_norm": 880.1036987304688,
      "learning_rate": 9.187486070227938e-06,
      "loss": 124.1478,
      "step": 32870
    },
    {
      "epoch": 0.2719940439260454,
      "grad_norm": 875.8194580078125,
      "learning_rate": 9.186715220416463e-06,
      "loss": 80.9601,
      "step": 32880
    },
    {
      "epoch": 0.2720767671754146,
      "grad_norm": 517.209716796875,
      "learning_rate": 9.185944037487271e-06,
      "loss": 105.6458,
      "step": 32890
    },
    {
      "epoch": 0.2721594904247839,
      "grad_norm": 783.9718627929688,
      "learning_rate": 9.185172521501723e-06,
      "loss": 103.928,
      "step": 32900
    },
    {
      "epoch": 0.27224221367415313,
      "grad_norm": 809.2305297851562,
      "learning_rate": 9.184400672521204e-06,
      "loss": 119.6438,
      "step": 32910
    },
    {
      "epoch": 0.27232493692352233,
      "grad_norm": 847.0447998046875,
      "learning_rate": 9.183628490607129e-06,
      "loss": 118.6409,
      "step": 32920
    },
    {
      "epoch": 0.2724076601728916,
      "grad_norm": 803.6466674804688,
      "learning_rate": 9.182855975820934e-06,
      "loss": 86.8706,
      "step": 32930
    },
    {
      "epoch": 0.27249038342226084,
      "grad_norm": 1112.55322265625,
      "learning_rate": 9.182083128224086e-06,
      "loss": 108.3938,
      "step": 32940
    },
    {
      "epoch": 0.27257310667163004,
      "grad_norm": 758.2708740234375,
      "learning_rate": 9.181309947878077e-06,
      "loss": 136.1542,
      "step": 32950
    },
    {
      "epoch": 0.2726558299209993,
      "grad_norm": 655.5614013671875,
      "learning_rate": 9.180536434844426e-06,
      "loss": 93.3358,
      "step": 32960
    },
    {
      "epoch": 0.27273855317036855,
      "grad_norm": 746.388671875,
      "learning_rate": 9.179762589184676e-06,
      "loss": 137.545,
      "step": 32970
    },
    {
      "epoch": 0.27282127641973775,
      "grad_norm": 919.367431640625,
      "learning_rate": 9.1789884109604e-06,
      "loss": 105.9232,
      "step": 32980
    },
    {
      "epoch": 0.272903999669107,
      "grad_norm": 1165.487548828125,
      "learning_rate": 9.178213900233193e-06,
      "loss": 119.8975,
      "step": 32990
    },
    {
      "epoch": 0.27298672291847625,
      "grad_norm": 1274.4398193359375,
      "learning_rate": 9.177439057064684e-06,
      "loss": 108.2796,
      "step": 33000
    },
    {
      "epoch": 0.27306944616784545,
      "grad_norm": 2099.3134765625,
      "learning_rate": 9.17666388151652e-06,
      "loss": 88.9919,
      "step": 33010
    },
    {
      "epoch": 0.2731521694172147,
      "grad_norm": 1373.3428955078125,
      "learning_rate": 9.175888373650377e-06,
      "loss": 109.9396,
      "step": 33020
    },
    {
      "epoch": 0.27323489266658396,
      "grad_norm": 1429.390869140625,
      "learning_rate": 9.175112533527963e-06,
      "loss": 94.565,
      "step": 33030
    },
    {
      "epoch": 0.27331761591595316,
      "grad_norm": 1118.853271484375,
      "learning_rate": 9.174336361211007e-06,
      "loss": 101.014,
      "step": 33040
    },
    {
      "epoch": 0.2734003391653224,
      "grad_norm": 2152.85107421875,
      "learning_rate": 9.173559856761262e-06,
      "loss": 153.9467,
      "step": 33050
    },
    {
      "epoch": 0.27348306241469167,
      "grad_norm": 909.071533203125,
      "learning_rate": 9.172783020240514e-06,
      "loss": 102.0454,
      "step": 33060
    },
    {
      "epoch": 0.27356578566406087,
      "grad_norm": 890.9006958007812,
      "learning_rate": 9.172005851710573e-06,
      "loss": 130.9717,
      "step": 33070
    },
    {
      "epoch": 0.2736485089134301,
      "grad_norm": 1000.61279296875,
      "learning_rate": 9.171228351233272e-06,
      "loss": 150.3027,
      "step": 33080
    },
    {
      "epoch": 0.2737312321627994,
      "grad_norm": 1430.8470458984375,
      "learning_rate": 9.170450518870475e-06,
      "loss": 149.3742,
      "step": 33090
    },
    {
      "epoch": 0.2738139554121686,
      "grad_norm": 1026.9654541015625,
      "learning_rate": 9.169672354684069e-06,
      "loss": 123.5882,
      "step": 33100
    },
    {
      "epoch": 0.27389667866153783,
      "grad_norm": 841.4974975585938,
      "learning_rate": 9.168893858735972e-06,
      "loss": 92.7002,
      "step": 33110
    },
    {
      "epoch": 0.2739794019109071,
      "grad_norm": 1173.48681640625,
      "learning_rate": 9.168115031088122e-06,
      "loss": 89.6682,
      "step": 33120
    },
    {
      "epoch": 0.2740621251602763,
      "grad_norm": 733.5807495117188,
      "learning_rate": 9.167335871802488e-06,
      "loss": 86.3547,
      "step": 33130
    },
    {
      "epoch": 0.27414484840964554,
      "grad_norm": 1434.056640625,
      "learning_rate": 9.166556380941063e-06,
      "loss": 125.5328,
      "step": 33140
    },
    {
      "epoch": 0.2742275716590148,
      "grad_norm": 820.1240844726562,
      "learning_rate": 9.16577655856587e-06,
      "loss": 130.4958,
      "step": 33150
    },
    {
      "epoch": 0.274310294908384,
      "grad_norm": 840.7511596679688,
      "learning_rate": 9.164996404738955e-06,
      "loss": 160.6511,
      "step": 33160
    },
    {
      "epoch": 0.27439301815775324,
      "grad_norm": 915.5698852539062,
      "learning_rate": 9.16421591952239e-06,
      "loss": 127.9161,
      "step": 33170
    },
    {
      "epoch": 0.2744757414071225,
      "grad_norm": 1336.1126708984375,
      "learning_rate": 9.163435102978276e-06,
      "loss": 122.7304,
      "step": 33180
    },
    {
      "epoch": 0.2745584646564917,
      "grad_norm": 982.8937377929688,
      "learning_rate": 9.162653955168739e-06,
      "loss": 118.9783,
      "step": 33190
    },
    {
      "epoch": 0.27464118790586095,
      "grad_norm": 756.2153930664062,
      "learning_rate": 9.161872476155929e-06,
      "loss": 101.5269,
      "step": 33200
    },
    {
      "epoch": 0.2747239111552302,
      "grad_norm": 732.1403198242188,
      "learning_rate": 9.161090666002029e-06,
      "loss": 124.4535,
      "step": 33210
    },
    {
      "epoch": 0.2748066344045994,
      "grad_norm": 753.6777954101562,
      "learning_rate": 9.16030852476924e-06,
      "loss": 109.3473,
      "step": 33220
    },
    {
      "epoch": 0.27488935765396866,
      "grad_norm": 822.023681640625,
      "learning_rate": 9.159526052519794e-06,
      "loss": 120.7444,
      "step": 33230
    },
    {
      "epoch": 0.27497208090333786,
      "grad_norm": 925.0752563476562,
      "learning_rate": 9.15874324931595e-06,
      "loss": 100.8344,
      "step": 33240
    },
    {
      "epoch": 0.2750548041527071,
      "grad_norm": 686.9509887695312,
      "learning_rate": 9.157960115219993e-06,
      "loss": 113.704,
      "step": 33250
    },
    {
      "epoch": 0.27513752740207637,
      "grad_norm": 996.56787109375,
      "learning_rate": 9.157176650294231e-06,
      "loss": 133.5279,
      "step": 33260
    },
    {
      "epoch": 0.27522025065144556,
      "grad_norm": 728.7750244140625,
      "learning_rate": 9.156392854601001e-06,
      "loss": 133.3526,
      "step": 33270
    },
    {
      "epoch": 0.2753029739008148,
      "grad_norm": 1302.504638671875,
      "learning_rate": 9.155608728202669e-06,
      "loss": 113.5402,
      "step": 33280
    },
    {
      "epoch": 0.2753856971501841,
      "grad_norm": 857.5109252929688,
      "learning_rate": 9.154824271161621e-06,
      "loss": 83.4826,
      "step": 33290
    },
    {
      "epoch": 0.27546842039955327,
      "grad_norm": 910.7625122070312,
      "learning_rate": 9.154039483540273e-06,
      "loss": 110.0397,
      "step": 33300
    },
    {
      "epoch": 0.2755511436489225,
      "grad_norm": 529.6849365234375,
      "learning_rate": 9.153254365401069e-06,
      "loss": 125.4888,
      "step": 33310
    },
    {
      "epoch": 0.2756338668982918,
      "grad_norm": 879.7425537109375,
      "learning_rate": 9.152468916806477e-06,
      "loss": 103.2796,
      "step": 33320
    },
    {
      "epoch": 0.275716590147661,
      "grad_norm": 1299.931884765625,
      "learning_rate": 9.151683137818989e-06,
      "loss": 119.4664,
      "step": 33330
    },
    {
      "epoch": 0.27579931339703023,
      "grad_norm": 698.8023681640625,
      "learning_rate": 9.150897028501126e-06,
      "loss": 100.8363,
      "step": 33340
    },
    {
      "epoch": 0.2758820366463995,
      "grad_norm": 1195.03466796875,
      "learning_rate": 9.15011058891544e-06,
      "loss": 119.996,
      "step": 33350
    },
    {
      "epoch": 0.2759647598957687,
      "grad_norm": 1072.2655029296875,
      "learning_rate": 9.149323819124498e-06,
      "loss": 113.2403,
      "step": 33360
    },
    {
      "epoch": 0.27604748314513794,
      "grad_norm": 897.378662109375,
      "learning_rate": 9.148536719190904e-06,
      "loss": 131.1827,
      "step": 33370
    },
    {
      "epoch": 0.2761302063945072,
      "grad_norm": 900.53955078125,
      "learning_rate": 9.147749289177282e-06,
      "loss": 141.6734,
      "step": 33380
    },
    {
      "epoch": 0.2762129296438764,
      "grad_norm": 809.6741333007812,
      "learning_rate": 9.146961529146285e-06,
      "loss": 99.829,
      "step": 33390
    },
    {
      "epoch": 0.27629565289324565,
      "grad_norm": 757.1586303710938,
      "learning_rate": 9.146173439160591e-06,
      "loss": 117.2545,
      "step": 33400
    },
    {
      "epoch": 0.2763783761426149,
      "grad_norm": 763.6544799804688,
      "learning_rate": 9.145385019282904e-06,
      "loss": 135.9243,
      "step": 33410
    },
    {
      "epoch": 0.2764610993919841,
      "grad_norm": 1427.904296875,
      "learning_rate": 9.144596269575957e-06,
      "loss": 125.229,
      "step": 33420
    },
    {
      "epoch": 0.27654382264135335,
      "grad_norm": 1012.0225219726562,
      "learning_rate": 9.143807190102504e-06,
      "loss": 126.2279,
      "step": 33430
    },
    {
      "epoch": 0.2766265458907226,
      "grad_norm": 2056.858154296875,
      "learning_rate": 9.143017780925331e-06,
      "loss": 153.6504,
      "step": 33440
    },
    {
      "epoch": 0.2767092691400918,
      "grad_norm": 1273.417236328125,
      "learning_rate": 9.142228042107248e-06,
      "loss": 109.7093,
      "step": 33450
    },
    {
      "epoch": 0.27679199238946106,
      "grad_norm": 1077.05126953125,
      "learning_rate": 9.141437973711092e-06,
      "loss": 106.108,
      "step": 33460
    },
    {
      "epoch": 0.2768747156388303,
      "grad_norm": 1240.419189453125,
      "learning_rate": 9.14064757579972e-06,
      "loss": 95.5216,
      "step": 33470
    },
    {
      "epoch": 0.2769574388881995,
      "grad_norm": 1002.42724609375,
      "learning_rate": 9.139856848436023e-06,
      "loss": 117.1653,
      "step": 33480
    },
    {
      "epoch": 0.27704016213756877,
      "grad_norm": 1525.692626953125,
      "learning_rate": 9.139065791682916e-06,
      "loss": 161.3095,
      "step": 33490
    },
    {
      "epoch": 0.277122885386938,
      "grad_norm": 773.1032104492188,
      "learning_rate": 9.138274405603342e-06,
      "loss": 168.1776,
      "step": 33500
    },
    {
      "epoch": 0.2772056086363072,
      "grad_norm": 768.6693115234375,
      "learning_rate": 9.137482690260265e-06,
      "loss": 156.8843,
      "step": 33510
    },
    {
      "epoch": 0.2772883318856765,
      "grad_norm": 307.3286437988281,
      "learning_rate": 9.13669064571668e-06,
      "loss": 109.4442,
      "step": 33520
    },
    {
      "epoch": 0.27737105513504573,
      "grad_norm": 3717.694580078125,
      "learning_rate": 9.135898272035601e-06,
      "loss": 167.7826,
      "step": 33530
    },
    {
      "epoch": 0.27745377838441493,
      "grad_norm": 986.0308227539062,
      "learning_rate": 9.13510556928008e-06,
      "loss": 87.5478,
      "step": 33540
    },
    {
      "epoch": 0.2775365016337842,
      "grad_norm": 652.1424560546875,
      "learning_rate": 9.134312537513188e-06,
      "loss": 106.2764,
      "step": 33550
    },
    {
      "epoch": 0.27761922488315344,
      "grad_norm": 720.5858154296875,
      "learning_rate": 9.133519176798021e-06,
      "loss": 152.1906,
      "step": 33560
    },
    {
      "epoch": 0.27770194813252264,
      "grad_norm": 1833.9097900390625,
      "learning_rate": 9.132725487197701e-06,
      "loss": 116.5092,
      "step": 33570
    },
    {
      "epoch": 0.2777846713818919,
      "grad_norm": 1374.428955078125,
      "learning_rate": 9.131931468775382e-06,
      "loss": 132.6865,
      "step": 33580
    },
    {
      "epoch": 0.2778673946312611,
      "grad_norm": 724.4324340820312,
      "learning_rate": 9.131137121594239e-06,
      "loss": 127.3931,
      "step": 33590
    },
    {
      "epoch": 0.27795011788063034,
      "grad_norm": 594.5848999023438,
      "learning_rate": 9.130342445717474e-06,
      "loss": 117.9793,
      "step": 33600
    },
    {
      "epoch": 0.2780328411299996,
      "grad_norm": 1347.4599609375,
      "learning_rate": 9.129547441208317e-06,
      "loss": 123.5553,
      "step": 33610
    },
    {
      "epoch": 0.2781155643793688,
      "grad_norm": 825.2183837890625,
      "learning_rate": 9.128752108130022e-06,
      "loss": 109.409,
      "step": 33620
    },
    {
      "epoch": 0.27819828762873805,
      "grad_norm": 952.2738037109375,
      "learning_rate": 9.12795644654587e-06,
      "loss": 149.2746,
      "step": 33630
    },
    {
      "epoch": 0.2782810108781073,
      "grad_norm": 6804.1904296875,
      "learning_rate": 9.127160456519168e-06,
      "loss": 132.9842,
      "step": 33640
    },
    {
      "epoch": 0.2783637341274765,
      "grad_norm": 953.0177612304688,
      "learning_rate": 9.126364138113251e-06,
      "loss": 119.1077,
      "step": 33650
    },
    {
      "epoch": 0.27844645737684576,
      "grad_norm": 771.2464599609375,
      "learning_rate": 9.125567491391476e-06,
      "loss": 117.123,
      "step": 33660
    },
    {
      "epoch": 0.278529180626215,
      "grad_norm": 1822.6439208984375,
      "learning_rate": 9.12477051641723e-06,
      "loss": 151.1186,
      "step": 33670
    },
    {
      "epoch": 0.2786119038755842,
      "grad_norm": 872.4586791992188,
      "learning_rate": 9.123973213253923e-06,
      "loss": 112.0873,
      "step": 33680
    },
    {
      "epoch": 0.27869462712495346,
      "grad_norm": 494.63134765625,
      "learning_rate": 9.123175581964995e-06,
      "loss": 100.9059,
      "step": 33690
    },
    {
      "epoch": 0.2787773503743227,
      "grad_norm": 959.332275390625,
      "learning_rate": 9.122377622613909e-06,
      "loss": 106.5335,
      "step": 33700
    },
    {
      "epoch": 0.2788600736236919,
      "grad_norm": 886.007568359375,
      "learning_rate": 9.121579335264155e-06,
      "loss": 128.5588,
      "step": 33710
    },
    {
      "epoch": 0.27894279687306117,
      "grad_norm": 1174.72265625,
      "learning_rate": 9.120780719979248e-06,
      "loss": 92.5889,
      "step": 33720
    },
    {
      "epoch": 0.2790255201224304,
      "grad_norm": 2207.4111328125,
      "learning_rate": 9.11998177682273e-06,
      "loss": 132.5646,
      "step": 33730
    },
    {
      "epoch": 0.2791082433717996,
      "grad_norm": 730.7549438476562,
      "learning_rate": 9.11918250585817e-06,
      "loss": 82.5129,
      "step": 33740
    },
    {
      "epoch": 0.2791909666211689,
      "grad_norm": 876.688232421875,
      "learning_rate": 9.118382907149164e-06,
      "loss": 109.9412,
      "step": 33750
    },
    {
      "epoch": 0.27927368987053813,
      "grad_norm": 831.474609375,
      "learning_rate": 9.117582980759332e-06,
      "loss": 124.5468,
      "step": 33760
    },
    {
      "epoch": 0.27935641311990733,
      "grad_norm": 820.8474731445312,
      "learning_rate": 9.116782726752317e-06,
      "loss": 126.4644,
      "step": 33770
    },
    {
      "epoch": 0.2794391363692766,
      "grad_norm": 921.0737915039062,
      "learning_rate": 9.115982145191796e-06,
      "loss": 116.8273,
      "step": 33780
    },
    {
      "epoch": 0.27952185961864584,
      "grad_norm": 1426.6785888671875,
      "learning_rate": 9.115181236141463e-06,
      "loss": 127.0457,
      "step": 33790
    },
    {
      "epoch": 0.27960458286801504,
      "grad_norm": 1093.164794921875,
      "learning_rate": 9.114379999665047e-06,
      "loss": 85.9147,
      "step": 33800
    },
    {
      "epoch": 0.2796873061173843,
      "grad_norm": 1163.352783203125,
      "learning_rate": 9.113578435826295e-06,
      "loss": 140.9147,
      "step": 33810
    },
    {
      "epoch": 0.27977002936675355,
      "grad_norm": 412.13787841796875,
      "learning_rate": 9.112776544688988e-06,
      "loss": 127.3203,
      "step": 33820
    },
    {
      "epoch": 0.27985275261612275,
      "grad_norm": 785.0323486328125,
      "learning_rate": 9.111974326316926e-06,
      "loss": 103.5417,
      "step": 33830
    },
    {
      "epoch": 0.279935475865492,
      "grad_norm": 1045.812255859375,
      "learning_rate": 9.111171780773938e-06,
      "loss": 120.1187,
      "step": 33840
    },
    {
      "epoch": 0.28001819911486125,
      "grad_norm": 916.0906372070312,
      "learning_rate": 9.110368908123878e-06,
      "loss": 139.0241,
      "step": 33850
    },
    {
      "epoch": 0.28010092236423045,
      "grad_norm": 1398.31982421875,
      "learning_rate": 9.10956570843063e-06,
      "loss": 132.5252,
      "step": 33860
    },
    {
      "epoch": 0.2801836456135997,
      "grad_norm": 1527.2587890625,
      "learning_rate": 9.108762181758096e-06,
      "loss": 130.4855,
      "step": 33870
    },
    {
      "epoch": 0.28026636886296896,
      "grad_norm": 751.3243408203125,
      "learning_rate": 9.107958328170215e-06,
      "loss": 106.5942,
      "step": 33880
    },
    {
      "epoch": 0.28034909211233816,
      "grad_norm": 1024.8853759765625,
      "learning_rate": 9.10715414773094e-06,
      "loss": 109.6669,
      "step": 33890
    },
    {
      "epoch": 0.2804318153617074,
      "grad_norm": 1099.5294189453125,
      "learning_rate": 9.10634964050426e-06,
      "loss": 119.2499,
      "step": 33900
    },
    {
      "epoch": 0.28051453861107667,
      "grad_norm": 775.4506225585938,
      "learning_rate": 9.105544806554184e-06,
      "loss": 157.9587,
      "step": 33910
    },
    {
      "epoch": 0.28059726186044587,
      "grad_norm": 1175.566162109375,
      "learning_rate": 9.104739645944752e-06,
      "loss": 179.6702,
      "step": 33920
    },
    {
      "epoch": 0.2806799851098151,
      "grad_norm": 1368.4671630859375,
      "learning_rate": 9.103934158740023e-06,
      "loss": 129.1513,
      "step": 33930
    },
    {
      "epoch": 0.2807627083591844,
      "grad_norm": 1385.238037109375,
      "learning_rate": 9.10312834500409e-06,
      "loss": 150.3397,
      "step": 33940
    },
    {
      "epoch": 0.2808454316085536,
      "grad_norm": 1050.946044921875,
      "learning_rate": 9.102322204801062e-06,
      "loss": 118.2614,
      "step": 33950
    },
    {
      "epoch": 0.28092815485792283,
      "grad_norm": 2881.64013671875,
      "learning_rate": 9.101515738195084e-06,
      "loss": 100.6495,
      "step": 33960
    },
    {
      "epoch": 0.28101087810729203,
      "grad_norm": 394.7975769042969,
      "learning_rate": 9.100708945250322e-06,
      "loss": 81.3734,
      "step": 33970
    },
    {
      "epoch": 0.2810936013566613,
      "grad_norm": 1190.456298828125,
      "learning_rate": 9.099901826030969e-06,
      "loss": 130.64,
      "step": 33980
    },
    {
      "epoch": 0.28117632460603054,
      "grad_norm": 1374.1256103515625,
      "learning_rate": 9.099094380601244e-06,
      "loss": 119.4305,
      "step": 33990
    },
    {
      "epoch": 0.28125904785539974,
      "grad_norm": 1412.7530517578125,
      "learning_rate": 9.098286609025392e-06,
      "loss": 106.0938,
      "step": 34000
    },
    {
      "epoch": 0.281341771104769,
      "grad_norm": 617.6422119140625,
      "learning_rate": 9.097478511367682e-06,
      "loss": 119.852,
      "step": 34010
    },
    {
      "epoch": 0.28142449435413824,
      "grad_norm": 857.2781982421875,
      "learning_rate": 9.096670087692413e-06,
      "loss": 168.8287,
      "step": 34020
    },
    {
      "epoch": 0.28150721760350744,
      "grad_norm": 0.0,
      "learning_rate": 9.095861338063906e-06,
      "loss": 99.5146,
      "step": 34030
    },
    {
      "epoch": 0.2815899408528767,
      "grad_norm": 1164.4888916015625,
      "learning_rate": 9.09505226254651e-06,
      "loss": 139.5455,
      "step": 34040
    },
    {
      "epoch": 0.28167266410224595,
      "grad_norm": 1813.4522705078125,
      "learning_rate": 9.094242861204598e-06,
      "loss": 153.4502,
      "step": 34050
    },
    {
      "epoch": 0.28175538735161515,
      "grad_norm": 918.6724243164062,
      "learning_rate": 9.093433134102572e-06,
      "loss": 95.0513,
      "step": 34060
    },
    {
      "epoch": 0.2818381106009844,
      "grad_norm": 758.8914794921875,
      "learning_rate": 9.09262308130486e-06,
      "loss": 127.6234,
      "step": 34070
    },
    {
      "epoch": 0.28192083385035366,
      "grad_norm": 1214.897705078125,
      "learning_rate": 9.091812702875908e-06,
      "loss": 135.7131,
      "step": 34080
    },
    {
      "epoch": 0.28200355709972286,
      "grad_norm": 1762.4208984375,
      "learning_rate": 9.0910019988802e-06,
      "loss": 116.575,
      "step": 34090
    },
    {
      "epoch": 0.2820862803490921,
      "grad_norm": 1027.57958984375,
      "learning_rate": 9.09019096938224e-06,
      "loss": 115.0406,
      "step": 34100
    },
    {
      "epoch": 0.28216900359846137,
      "grad_norm": 549.1436157226562,
      "learning_rate": 9.089379614446554e-06,
      "loss": 101.1465,
      "step": 34110
    },
    {
      "epoch": 0.28225172684783056,
      "grad_norm": 1149.4486083984375,
      "learning_rate": 9.0885679341377e-06,
      "loss": 111.7543,
      "step": 34120
    },
    {
      "epoch": 0.2823344500971998,
      "grad_norm": 827.5548095703125,
      "learning_rate": 9.08775592852026e-06,
      "loss": 84.8386,
      "step": 34130
    },
    {
      "epoch": 0.2824171733465691,
      "grad_norm": 2282.6884765625,
      "learning_rate": 9.08694359765884e-06,
      "loss": 122.2059,
      "step": 34140
    },
    {
      "epoch": 0.28249989659593827,
      "grad_norm": 1563.5760498046875,
      "learning_rate": 9.086130941618075e-06,
      "loss": 127.1692,
      "step": 34150
    },
    {
      "epoch": 0.2825826198453075,
      "grad_norm": 546.21337890625,
      "learning_rate": 9.085317960462625e-06,
      "loss": 91.0043,
      "step": 34160
    },
    {
      "epoch": 0.2826653430946768,
      "grad_norm": 1092.4716796875,
      "learning_rate": 9.084504654257173e-06,
      "loss": 126.0462,
      "step": 34170
    },
    {
      "epoch": 0.282748066344046,
      "grad_norm": 1006.257568359375,
      "learning_rate": 9.08369102306643e-06,
      "loss": 104.315,
      "step": 34180
    },
    {
      "epoch": 0.28283078959341523,
      "grad_norm": 1649.3040771484375,
      "learning_rate": 9.082877066955135e-06,
      "loss": 101.4608,
      "step": 34190
    },
    {
      "epoch": 0.2829135128427845,
      "grad_norm": 1023.9491577148438,
      "learning_rate": 9.08206278598805e-06,
      "loss": 133.0885,
      "step": 34200
    },
    {
      "epoch": 0.2829962360921537,
      "grad_norm": 1013.92041015625,
      "learning_rate": 9.081248180229963e-06,
      "loss": 93.8945,
      "step": 34210
    },
    {
      "epoch": 0.28307895934152294,
      "grad_norm": 655.7772827148438,
      "learning_rate": 9.080433249745688e-06,
      "loss": 104.3141,
      "step": 34220
    },
    {
      "epoch": 0.2831616825908922,
      "grad_norm": 1191.8131103515625,
      "learning_rate": 9.079617994600066e-06,
      "loss": 147.7726,
      "step": 34230
    },
    {
      "epoch": 0.2832444058402614,
      "grad_norm": 628.782470703125,
      "learning_rate": 9.078802414857963e-06,
      "loss": 96.9786,
      "step": 34240
    },
    {
      "epoch": 0.28332712908963065,
      "grad_norm": 695.0119018554688,
      "learning_rate": 9.077986510584273e-06,
      "loss": 111.2695,
      "step": 34250
    },
    {
      "epoch": 0.2834098523389999,
      "grad_norm": 1148.8243408203125,
      "learning_rate": 9.07717028184391e-06,
      "loss": 104.8876,
      "step": 34260
    },
    {
      "epoch": 0.2834925755883691,
      "grad_norm": 766.49072265625,
      "learning_rate": 9.07635372870182e-06,
      "loss": 84.2609,
      "step": 34270
    },
    {
      "epoch": 0.28357529883773835,
      "grad_norm": 1150.672119140625,
      "learning_rate": 9.07553685122297e-06,
      "loss": 146.0944,
      "step": 34280
    },
    {
      "epoch": 0.2836580220871076,
      "grad_norm": 1328.8944091796875,
      "learning_rate": 9.074719649472358e-06,
      "loss": 115.0404,
      "step": 34290
    },
    {
      "epoch": 0.2837407453364768,
      "grad_norm": 1080.066650390625,
      "learning_rate": 9.073902123515005e-06,
      "loss": 144.9194,
      "step": 34300
    },
    {
      "epoch": 0.28382346858584606,
      "grad_norm": 676.450439453125,
      "learning_rate": 9.073084273415956e-06,
      "loss": 79.431,
      "step": 34310
    },
    {
      "epoch": 0.28390619183521526,
      "grad_norm": 1445.7745361328125,
      "learning_rate": 9.072266099240286e-06,
      "loss": 99.8113,
      "step": 34320
    },
    {
      "epoch": 0.2839889150845845,
      "grad_norm": 739.0885009765625,
      "learning_rate": 9.07144760105309e-06,
      "loss": 142.6431,
      "step": 34330
    },
    {
      "epoch": 0.28407163833395377,
      "grad_norm": 920.0739135742188,
      "learning_rate": 9.070628778919493e-06,
      "loss": 93.1577,
      "step": 34340
    },
    {
      "epoch": 0.28415436158332297,
      "grad_norm": 1188.843017578125,
      "learning_rate": 9.069809632904647e-06,
      "loss": 111.0898,
      "step": 34350
    },
    {
      "epoch": 0.2842370848326922,
      "grad_norm": 991.0109252929688,
      "learning_rate": 9.068990163073726e-06,
      "loss": 87.4462,
      "step": 34360
    },
    {
      "epoch": 0.2843198080820615,
      "grad_norm": 845.0187377929688,
      "learning_rate": 9.068170369491932e-06,
      "loss": 95.4965,
      "step": 34370
    },
    {
      "epoch": 0.2844025313314307,
      "grad_norm": 861.9247436523438,
      "learning_rate": 9.067350252224491e-06,
      "loss": 76.6258,
      "step": 34380
    },
    {
      "epoch": 0.28448525458079993,
      "grad_norm": 1371.8582763671875,
      "learning_rate": 9.066529811336658e-06,
      "loss": 117.6796,
      "step": 34390
    },
    {
      "epoch": 0.2845679778301692,
      "grad_norm": 909.2398681640625,
      "learning_rate": 9.06570904689371e-06,
      "loss": 96.8476,
      "step": 34400
    },
    {
      "epoch": 0.2846507010795384,
      "grad_norm": 941.552978515625,
      "learning_rate": 9.064887958960953e-06,
      "loss": 89.4982,
      "step": 34410
    },
    {
      "epoch": 0.28473342432890764,
      "grad_norm": 2689.89404296875,
      "learning_rate": 9.064066547603716e-06,
      "loss": 116.0555,
      "step": 34420
    },
    {
      "epoch": 0.2848161475782769,
      "grad_norm": 1277.172607421875,
      "learning_rate": 9.063244812887357e-06,
      "loss": 111.9663,
      "step": 34430
    },
    {
      "epoch": 0.2848988708276461,
      "grad_norm": 872.6428833007812,
      "learning_rate": 9.062422754877253e-06,
      "loss": 124.443,
      "step": 34440
    },
    {
      "epoch": 0.28498159407701534,
      "grad_norm": 698.310546875,
      "learning_rate": 9.061600373638816e-06,
      "loss": 127.376,
      "step": 34450
    },
    {
      "epoch": 0.2850643173263846,
      "grad_norm": 914.8101806640625,
      "learning_rate": 9.06077766923748e-06,
      "loss": 113.9157,
      "step": 34460
    },
    {
      "epoch": 0.2851470405757538,
      "grad_norm": 1122.4351806640625,
      "learning_rate": 9.059954641738697e-06,
      "loss": 126.3374,
      "step": 34470
    },
    {
      "epoch": 0.28522976382512305,
      "grad_norm": 1123.8736572265625,
      "learning_rate": 9.059131291207958e-06,
      "loss": 105.3611,
      "step": 34480
    },
    {
      "epoch": 0.2853124870744923,
      "grad_norm": 656.4722290039062,
      "learning_rate": 9.058307617710771e-06,
      "loss": 142.5772,
      "step": 34490
    },
    {
      "epoch": 0.2853952103238615,
      "grad_norm": 608.4116821289062,
      "learning_rate": 9.057483621312671e-06,
      "loss": 115.3732,
      "step": 34500
    },
    {
      "epoch": 0.28547793357323076,
      "grad_norm": 581.153564453125,
      "learning_rate": 9.056659302079222e-06,
      "loss": 118.1142,
      "step": 34510
    },
    {
      "epoch": 0.2855606568226,
      "grad_norm": 1226.16552734375,
      "learning_rate": 9.055834660076008e-06,
      "loss": 110.4029,
      "step": 34520
    },
    {
      "epoch": 0.2856433800719692,
      "grad_norm": 1013.7431640625,
      "learning_rate": 9.055009695368646e-06,
      "loss": 135.0033,
      "step": 34530
    },
    {
      "epoch": 0.28572610332133846,
      "grad_norm": 749.7244262695312,
      "learning_rate": 9.054184408022772e-06,
      "loss": 157.7157,
      "step": 34540
    },
    {
      "epoch": 0.2858088265707077,
      "grad_norm": 456.33026123046875,
      "learning_rate": 9.05335879810405e-06,
      "loss": 145.5047,
      "step": 34550
    },
    {
      "epoch": 0.2858915498200769,
      "grad_norm": 622.3046264648438,
      "learning_rate": 9.052532865678171e-06,
      "loss": 123.2549,
      "step": 34560
    },
    {
      "epoch": 0.28597427306944617,
      "grad_norm": 1478.7266845703125,
      "learning_rate": 9.05170661081085e-06,
      "loss": 127.2245,
      "step": 34570
    },
    {
      "epoch": 0.2860569963188154,
      "grad_norm": 619.5127563476562,
      "learning_rate": 9.050880033567831e-06,
      "loss": 132.0401,
      "step": 34580
    },
    {
      "epoch": 0.2861397195681846,
      "grad_norm": 1351.7064208984375,
      "learning_rate": 9.050053134014878e-06,
      "loss": 100.6227,
      "step": 34590
    },
    {
      "epoch": 0.2862224428175539,
      "grad_norm": 1285.1185302734375,
      "learning_rate": 9.049225912217782e-06,
      "loss": 102.2231,
      "step": 34600
    },
    {
      "epoch": 0.28630516606692313,
      "grad_norm": 1479.285888671875,
      "learning_rate": 9.048398368242365e-06,
      "loss": 93.523,
      "step": 34610
    },
    {
      "epoch": 0.28638788931629233,
      "grad_norm": 845.21826171875,
      "learning_rate": 9.047570502154471e-06,
      "loss": 97.2673,
      "step": 34620
    },
    {
      "epoch": 0.2864706125656616,
      "grad_norm": 1318.1756591796875,
      "learning_rate": 9.046742314019968e-06,
      "loss": 136.1264,
      "step": 34630
    },
    {
      "epoch": 0.28655333581503084,
      "grad_norm": 1689.4859619140625,
      "learning_rate": 9.045913803904748e-06,
      "loss": 127.8525,
      "step": 34640
    },
    {
      "epoch": 0.28663605906440004,
      "grad_norm": 759.2603149414062,
      "learning_rate": 9.045084971874738e-06,
      "loss": 119.6263,
      "step": 34650
    },
    {
      "epoch": 0.2867187823137693,
      "grad_norm": 1472.703857421875,
      "learning_rate": 9.04425581799588e-06,
      "loss": 122.0518,
      "step": 34660
    },
    {
      "epoch": 0.2868015055631385,
      "grad_norm": 903.3499755859375,
      "learning_rate": 9.043426342334147e-06,
      "loss": 104.4996,
      "step": 34670
    },
    {
      "epoch": 0.28688422881250775,
      "grad_norm": 726.6737670898438,
      "learning_rate": 9.042596544955538e-06,
      "loss": 93.7296,
      "step": 34680
    },
    {
      "epoch": 0.286966952061877,
      "grad_norm": 1467.8472900390625,
      "learning_rate": 9.041766425926073e-06,
      "loss": 109.8673,
      "step": 34690
    },
    {
      "epoch": 0.2870496753112462,
      "grad_norm": 1312.5286865234375,
      "learning_rate": 9.040935985311804e-06,
      "loss": 83.5798,
      "step": 34700
    },
    {
      "epoch": 0.28713239856061545,
      "grad_norm": 1040.613525390625,
      "learning_rate": 9.040105223178803e-06,
      "loss": 105.7686,
      "step": 34710
    },
    {
      "epoch": 0.2872151218099847,
      "grad_norm": 737.3902587890625,
      "learning_rate": 9.039274139593173e-06,
      "loss": 100.6217,
      "step": 34720
    },
    {
      "epoch": 0.2872978450593539,
      "grad_norm": 1900.810302734375,
      "learning_rate": 9.038442734621034e-06,
      "loss": 108.4988,
      "step": 34730
    },
    {
      "epoch": 0.28738056830872316,
      "grad_norm": 823.1213989257812,
      "learning_rate": 9.037611008328544e-06,
      "loss": 87.0847,
      "step": 34740
    },
    {
      "epoch": 0.2874632915580924,
      "grad_norm": 1027.710693359375,
      "learning_rate": 9.036778960781874e-06,
      "loss": 96.1623,
      "step": 34750
    },
    {
      "epoch": 0.2875460148074616,
      "grad_norm": 1012.5924682617188,
      "learning_rate": 9.03594659204723e-06,
      "loss": 99.4886,
      "step": 34760
    },
    {
      "epoch": 0.28762873805683087,
      "grad_norm": 1981.74609375,
      "learning_rate": 9.035113902190838e-06,
      "loss": 131.8146,
      "step": 34770
    },
    {
      "epoch": 0.2877114613062001,
      "grad_norm": 838.7272338867188,
      "learning_rate": 9.03428089127895e-06,
      "loss": 138.8772,
      "step": 34780
    },
    {
      "epoch": 0.2877941845555693,
      "grad_norm": 614.944091796875,
      "learning_rate": 9.033447559377847e-06,
      "loss": 102.8104,
      "step": 34790
    },
    {
      "epoch": 0.2878769078049386,
      "grad_norm": 1577.7645263671875,
      "learning_rate": 9.032613906553833e-06,
      "loss": 128.1048,
      "step": 34800
    },
    {
      "epoch": 0.28795963105430783,
      "grad_norm": 1021.2303466796875,
      "learning_rate": 9.031779932873238e-06,
      "loss": 107.6142,
      "step": 34810
    },
    {
      "epoch": 0.28804235430367703,
      "grad_norm": 644.7843017578125,
      "learning_rate": 9.030945638402415e-06,
      "loss": 103.4213,
      "step": 34820
    },
    {
      "epoch": 0.2881250775530463,
      "grad_norm": 873.067138671875,
      "learning_rate": 9.030111023207751e-06,
      "loss": 93.0126,
      "step": 34830
    },
    {
      "epoch": 0.28820780080241554,
      "grad_norm": 1150.903564453125,
      "learning_rate": 9.029276087355646e-06,
      "loss": 121.5422,
      "step": 34840
    },
    {
      "epoch": 0.28829052405178474,
      "grad_norm": 573.4194946289062,
      "learning_rate": 9.028440830912536e-06,
      "loss": 120.833,
      "step": 34850
    },
    {
      "epoch": 0.288373247301154,
      "grad_norm": 951.9305419921875,
      "learning_rate": 9.027605253944874e-06,
      "loss": 146.5991,
      "step": 34860
    },
    {
      "epoch": 0.28845597055052324,
      "grad_norm": 646.8169555664062,
      "learning_rate": 9.026769356519149e-06,
      "loss": 73.8264,
      "step": 34870
    },
    {
      "epoch": 0.28853869379989244,
      "grad_norm": 2599.574462890625,
      "learning_rate": 9.025933138701865e-06,
      "loss": 121.7481,
      "step": 34880
    },
    {
      "epoch": 0.2886214170492617,
      "grad_norm": 964.8515625,
      "learning_rate": 9.02509660055956e-06,
      "loss": 98.844,
      "step": 34890
    },
    {
      "epoch": 0.28870414029863095,
      "grad_norm": 665.4266357421875,
      "learning_rate": 9.02425974215879e-06,
      "loss": 126.3558,
      "step": 34900
    },
    {
      "epoch": 0.28878686354800015,
      "grad_norm": 747.7141723632812,
      "learning_rate": 9.02342256356614e-06,
      "loss": 94.1504,
      "step": 34910
    },
    {
      "epoch": 0.2888695867973694,
      "grad_norm": 1255.60302734375,
      "learning_rate": 9.022585064848222e-06,
      "loss": 101.1276,
      "step": 34920
    },
    {
      "epoch": 0.28895231004673866,
      "grad_norm": 1393.7789306640625,
      "learning_rate": 9.021747246071673e-06,
      "loss": 106.2699,
      "step": 34930
    },
    {
      "epoch": 0.28903503329610786,
      "grad_norm": 860.56689453125,
      "learning_rate": 9.020909107303152e-06,
      "loss": 124.7073,
      "step": 34940
    },
    {
      "epoch": 0.2891177565454771,
      "grad_norm": 982.3703002929688,
      "learning_rate": 9.020070648609347e-06,
      "loss": 112.9233,
      "step": 34950
    },
    {
      "epoch": 0.28920047979484637,
      "grad_norm": 1855.6322021484375,
      "learning_rate": 9.01923187005697e-06,
      "loss": 108.5488,
      "step": 34960
    },
    {
      "epoch": 0.28928320304421556,
      "grad_norm": 904.826416015625,
      "learning_rate": 9.018392771712758e-06,
      "loss": 125.191,
      "step": 34970
    },
    {
      "epoch": 0.2893659262935848,
      "grad_norm": 704.2764892578125,
      "learning_rate": 9.017553353643479e-06,
      "loss": 126.2956,
      "step": 34980
    },
    {
      "epoch": 0.2894486495429541,
      "grad_norm": 956.4945678710938,
      "learning_rate": 9.016713615915913e-06,
      "loss": 81.5747,
      "step": 34990
    },
    {
      "epoch": 0.28953137279232327,
      "grad_norm": 783.7576293945312,
      "learning_rate": 9.01587355859688e-06,
      "loss": 92.5522,
      "step": 35000
    },
    {
      "epoch": 0.2896140960416925,
      "grad_norm": 1208.607421875,
      "learning_rate": 9.015033181753219e-06,
      "loss": 106.08,
      "step": 35010
    },
    {
      "epoch": 0.2896968192910618,
      "grad_norm": 852.6314697265625,
      "learning_rate": 9.014192485451794e-06,
      "loss": 116.7134,
      "step": 35020
    },
    {
      "epoch": 0.289779542540431,
      "grad_norm": 1022.0242309570312,
      "learning_rate": 9.013351469759497e-06,
      "loss": 144.6814,
      "step": 35030
    },
    {
      "epoch": 0.28986226578980023,
      "grad_norm": 1243.4373779296875,
      "learning_rate": 9.01251013474324e-06,
      "loss": 104.1561,
      "step": 35040
    },
    {
      "epoch": 0.28994498903916943,
      "grad_norm": 960.918701171875,
      "learning_rate": 9.011668480469969e-06,
      "loss": 105.9182,
      "step": 35050
    },
    {
      "epoch": 0.2900277122885387,
      "grad_norm": 601.8880004882812,
      "learning_rate": 9.010826507006644e-06,
      "loss": 94.2774,
      "step": 35060
    },
    {
      "epoch": 0.29011043553790794,
      "grad_norm": 642.360107421875,
      "learning_rate": 9.009984214420265e-06,
      "loss": 142.8863,
      "step": 35070
    },
    {
      "epoch": 0.29019315878727714,
      "grad_norm": 1095.9666748046875,
      "learning_rate": 9.009141602777845e-06,
      "loss": 128.7609,
      "step": 35080
    },
    {
      "epoch": 0.2902758820366464,
      "grad_norm": 1230.556396484375,
      "learning_rate": 9.008298672146425e-06,
      "loss": 111.0792,
      "step": 35090
    },
    {
      "epoch": 0.29035860528601565,
      "grad_norm": 1034.6533203125,
      "learning_rate": 9.007455422593077e-06,
      "loss": 111.8609,
      "step": 35100
    },
    {
      "epoch": 0.29044132853538485,
      "grad_norm": 509.1571960449219,
      "learning_rate": 9.006611854184893e-06,
      "loss": 88.8575,
      "step": 35110
    },
    {
      "epoch": 0.2905240517847541,
      "grad_norm": 832.6845703125,
      "learning_rate": 9.00576796698899e-06,
      "loss": 113.1765,
      "step": 35120
    },
    {
      "epoch": 0.29060677503412335,
      "grad_norm": 1436.23388671875,
      "learning_rate": 9.004923761072515e-06,
      "loss": 103.5024,
      "step": 35130
    },
    {
      "epoch": 0.29068949828349255,
      "grad_norm": 1610.103759765625,
      "learning_rate": 9.004079236502636e-06,
      "loss": 113.4215,
      "step": 35140
    },
    {
      "epoch": 0.2907722215328618,
      "grad_norm": 881.2527465820312,
      "learning_rate": 9.00323439334655e-06,
      "loss": 123.1393,
      "step": 35150
    },
    {
      "epoch": 0.29085494478223106,
      "grad_norm": 734.708740234375,
      "learning_rate": 9.002389231671474e-06,
      "loss": 121.4382,
      "step": 35160
    },
    {
      "epoch": 0.29093766803160026,
      "grad_norm": 787.5501708984375,
      "learning_rate": 9.001543751544654e-06,
      "loss": 107.295,
      "step": 35170
    },
    {
      "epoch": 0.2910203912809695,
      "grad_norm": 1131.2176513671875,
      "learning_rate": 9.000697953033364e-06,
      "loss": 107.077,
      "step": 35180
    },
    {
      "epoch": 0.29110311453033877,
      "grad_norm": 1214.8603515625,
      "learning_rate": 8.999851836204901e-06,
      "loss": 103.6586,
      "step": 35190
    },
    {
      "epoch": 0.29118583777970797,
      "grad_norm": 895.4462280273438,
      "learning_rate": 8.99900540112658e-06,
      "loss": 83.9514,
      "step": 35200
    },
    {
      "epoch": 0.2912685610290772,
      "grad_norm": 543.6385498046875,
      "learning_rate": 8.998158647865753e-06,
      "loss": 107.0998,
      "step": 35210
    },
    {
      "epoch": 0.2913512842784465,
      "grad_norm": 1422.5147705078125,
      "learning_rate": 8.997311576489793e-06,
      "loss": 127.3802,
      "step": 35220
    },
    {
      "epoch": 0.2914340075278157,
      "grad_norm": 1031.476806640625,
      "learning_rate": 8.996464187066096e-06,
      "loss": 132.6045,
      "step": 35230
    },
    {
      "epoch": 0.29151673077718493,
      "grad_norm": 962.57568359375,
      "learning_rate": 8.995616479662084e-06,
      "loss": 76.9806,
      "step": 35240
    },
    {
      "epoch": 0.2915994540265542,
      "grad_norm": 888.2576293945312,
      "learning_rate": 8.994768454345207e-06,
      "loss": 88.927,
      "step": 35250
    },
    {
      "epoch": 0.2916821772759234,
      "grad_norm": 991.271484375,
      "learning_rate": 8.993920111182937e-06,
      "loss": 116.7842,
      "step": 35260
    },
    {
      "epoch": 0.29176490052529264,
      "grad_norm": 2302.41064453125,
      "learning_rate": 8.993071450242775e-06,
      "loss": 99.9801,
      "step": 35270
    },
    {
      "epoch": 0.2918476237746619,
      "grad_norm": 729.2933349609375,
      "learning_rate": 8.99222247159224e-06,
      "loss": 108.2754,
      "step": 35280
    },
    {
      "epoch": 0.2919303470240311,
      "grad_norm": 940.2220458984375,
      "learning_rate": 8.991373175298887e-06,
      "loss": 104.497,
      "step": 35290
    },
    {
      "epoch": 0.29201307027340034,
      "grad_norm": 893.2850952148438,
      "learning_rate": 8.99052356143029e-06,
      "loss": 104.8119,
      "step": 35300
    },
    {
      "epoch": 0.2920957935227696,
      "grad_norm": 1792.8408203125,
      "learning_rate": 8.989673630054044e-06,
      "loss": 134.8155,
      "step": 35310
    },
    {
      "epoch": 0.2921785167721388,
      "grad_norm": 823.2114868164062,
      "learning_rate": 8.988823381237778e-06,
      "loss": 91.5063,
      "step": 35320
    },
    {
      "epoch": 0.29226124002150805,
      "grad_norm": 1062.8973388671875,
      "learning_rate": 8.987972815049144e-06,
      "loss": 130.8569,
      "step": 35330
    },
    {
      "epoch": 0.2923439632708773,
      "grad_norm": 676.7745361328125,
      "learning_rate": 8.987121931555814e-06,
      "loss": 124.0575,
      "step": 35340
    },
    {
      "epoch": 0.2924266865202465,
      "grad_norm": 1002.80029296875,
      "learning_rate": 8.986270730825489e-06,
      "loss": 101.8287,
      "step": 35350
    },
    {
      "epoch": 0.29250940976961576,
      "grad_norm": 1145.8223876953125,
      "learning_rate": 8.985419212925898e-06,
      "loss": 110.8668,
      "step": 35360
    },
    {
      "epoch": 0.292592133018985,
      "grad_norm": 992.4735717773438,
      "learning_rate": 8.98456737792479e-06,
      "loss": 80.9995,
      "step": 35370
    },
    {
      "epoch": 0.2926748562683542,
      "grad_norm": 488.52618408203125,
      "learning_rate": 8.983715225889942e-06,
      "loss": 111.5234,
      "step": 35380
    },
    {
      "epoch": 0.29275757951772347,
      "grad_norm": 1136.89794921875,
      "learning_rate": 8.982862756889158e-06,
      "loss": 162.2252,
      "step": 35390
    },
    {
      "epoch": 0.29284030276709266,
      "grad_norm": 1113.3115234375,
      "learning_rate": 8.982009970990262e-06,
      "loss": 89.9033,
      "step": 35400
    },
    {
      "epoch": 0.2929230260164619,
      "grad_norm": 1392.5189208984375,
      "learning_rate": 8.98115686826111e-06,
      "loss": 131.7283,
      "step": 35410
    },
    {
      "epoch": 0.2930057492658312,
      "grad_norm": 617.0336303710938,
      "learning_rate": 8.980303448769574e-06,
      "loss": 99.9001,
      "step": 35420
    },
    {
      "epoch": 0.29308847251520037,
      "grad_norm": 2442.5810546875,
      "learning_rate": 8.979449712583562e-06,
      "loss": 112.8064,
      "step": 35430
    },
    {
      "epoch": 0.2931711957645696,
      "grad_norm": 544.6372680664062,
      "learning_rate": 8.978595659770997e-06,
      "loss": 109.6494,
      "step": 35440
    },
    {
      "epoch": 0.2932539190139389,
      "grad_norm": 860.6021118164062,
      "learning_rate": 8.977741290399836e-06,
      "loss": 106.3515,
      "step": 35450
    },
    {
      "epoch": 0.2933366422633081,
      "grad_norm": 1385.670654296875,
      "learning_rate": 8.976886604538055e-06,
      "loss": 117.2203,
      "step": 35460
    },
    {
      "epoch": 0.29341936551267733,
      "grad_norm": 653.8483276367188,
      "learning_rate": 8.976031602253661e-06,
      "loss": 91.5749,
      "step": 35470
    },
    {
      "epoch": 0.2935020887620466,
      "grad_norm": 707.4151000976562,
      "learning_rate": 8.975176283614677e-06,
      "loss": 137.304,
      "step": 35480
    },
    {
      "epoch": 0.2935848120114158,
      "grad_norm": 724.526123046875,
      "learning_rate": 8.97432064868916e-06,
      "loss": 117.3225,
      "step": 35490
    },
    {
      "epoch": 0.29366753526078504,
      "grad_norm": 684.9421997070312,
      "learning_rate": 8.973464697545191e-06,
      "loss": 118.9565,
      "step": 35500
    },
    {
      "epoch": 0.2937502585101543,
      "grad_norm": 1034.4615478515625,
      "learning_rate": 8.97260843025087e-06,
      "loss": 117.1274,
      "step": 35510
    },
    {
      "epoch": 0.2938329817595235,
      "grad_norm": 808.4039916992188,
      "learning_rate": 8.971751846874329e-06,
      "loss": 109.1277,
      "step": 35520
    },
    {
      "epoch": 0.29391570500889275,
      "grad_norm": 793.2233276367188,
      "learning_rate": 8.97089494748372e-06,
      "loss": 150.5428,
      "step": 35530
    },
    {
      "epoch": 0.293998428258262,
      "grad_norm": 1373.2613525390625,
      "learning_rate": 8.970037732147226e-06,
      "loss": 144.7835,
      "step": 35540
    },
    {
      "epoch": 0.2940811515076312,
      "grad_norm": 876.5359497070312,
      "learning_rate": 8.969180200933048e-06,
      "loss": 175.0752,
      "step": 35550
    },
    {
      "epoch": 0.29416387475700045,
      "grad_norm": 1215.04443359375,
      "learning_rate": 8.968322353909417e-06,
      "loss": 105.4971,
      "step": 35560
    },
    {
      "epoch": 0.2942465980063697,
      "grad_norm": 1001.4893188476562,
      "learning_rate": 8.96746419114459e-06,
      "loss": 109.2281,
      "step": 35570
    },
    {
      "epoch": 0.2943293212557389,
      "grad_norm": 1082.677001953125,
      "learning_rate": 8.966605712706844e-06,
      "loss": 90.3954,
      "step": 35580
    },
    {
      "epoch": 0.29441204450510816,
      "grad_norm": 962.15380859375,
      "learning_rate": 8.965746918664486e-06,
      "loss": 96.3644,
      "step": 35590
    },
    {
      "epoch": 0.2944947677544774,
      "grad_norm": 1004.3434448242188,
      "learning_rate": 8.964887809085846e-06,
      "loss": 128.4367,
      "step": 35600
    },
    {
      "epoch": 0.2945774910038466,
      "grad_norm": 535.0782470703125,
      "learning_rate": 8.96402838403928e-06,
      "loss": 101.0775,
      "step": 35610
    },
    {
      "epoch": 0.29466021425321587,
      "grad_norm": 451.95941162109375,
      "learning_rate": 8.96316864359317e-06,
      "loss": 103.9834,
      "step": 35620
    },
    {
      "epoch": 0.2947429375025851,
      "grad_norm": 861.1315307617188,
      "learning_rate": 8.962308587815916e-06,
      "loss": 95.3887,
      "step": 35630
    },
    {
      "epoch": 0.2948256607519543,
      "grad_norm": 873.6254272460938,
      "learning_rate": 8.961448216775955e-06,
      "loss": 106.6663,
      "step": 35640
    },
    {
      "epoch": 0.2949083840013236,
      "grad_norm": 647.5992431640625,
      "learning_rate": 8.960587530541737e-06,
      "loss": 113.9285,
      "step": 35650
    },
    {
      "epoch": 0.29499110725069283,
      "grad_norm": 818.12744140625,
      "learning_rate": 8.959726529181748e-06,
      "loss": 93.2586,
      "step": 35660
    },
    {
      "epoch": 0.29507383050006203,
      "grad_norm": 2233.892578125,
      "learning_rate": 8.95886521276449e-06,
      "loss": 99.4201,
      "step": 35670
    },
    {
      "epoch": 0.2951565537494313,
      "grad_norm": 1236.3218994140625,
      "learning_rate": 8.958003581358498e-06,
      "loss": 122.5037,
      "step": 35680
    },
    {
      "epoch": 0.29523927699880054,
      "grad_norm": 1387.99462890625,
      "learning_rate": 8.957141635032325e-06,
      "loss": 103.0061,
      "step": 35690
    },
    {
      "epoch": 0.29532200024816974,
      "grad_norm": 363.5679626464844,
      "learning_rate": 8.956279373854553e-06,
      "loss": 297.6621,
      "step": 35700
    },
    {
      "epoch": 0.295404723497539,
      "grad_norm": 848.06884765625,
      "learning_rate": 8.955416797893787e-06,
      "loss": 133.0075,
      "step": 35710
    },
    {
      "epoch": 0.29548744674690824,
      "grad_norm": 767.8993530273438,
      "learning_rate": 8.95455390721866e-06,
      "loss": 86.9913,
      "step": 35720
    },
    {
      "epoch": 0.29557016999627744,
      "grad_norm": 947.8824462890625,
      "learning_rate": 8.953690701897827e-06,
      "loss": 126.0984,
      "step": 35730
    },
    {
      "epoch": 0.2956528932456467,
      "grad_norm": 866.94970703125,
      "learning_rate": 8.952827181999973e-06,
      "loss": 100.0804,
      "step": 35740
    },
    {
      "epoch": 0.29573561649501595,
      "grad_norm": 997.9763793945312,
      "learning_rate": 8.951963347593797e-06,
      "loss": 158.5358,
      "step": 35750
    },
    {
      "epoch": 0.29581833974438515,
      "grad_norm": 682.8516235351562,
      "learning_rate": 8.951099198748036e-06,
      "loss": 101.2486,
      "step": 35760
    },
    {
      "epoch": 0.2959010629937544,
      "grad_norm": 804.0764770507812,
      "learning_rate": 8.950234735531445e-06,
      "loss": 98.5626,
      "step": 35770
    },
    {
      "epoch": 0.2959837862431236,
      "grad_norm": 10388.48046875,
      "learning_rate": 8.949369958012806e-06,
      "loss": 155.4089,
      "step": 35780
    },
    {
      "epoch": 0.29606650949249286,
      "grad_norm": 886.81298828125,
      "learning_rate": 8.948504866260924e-06,
      "loss": 113.0202,
      "step": 35790
    },
    {
      "epoch": 0.2961492327418621,
      "grad_norm": 897.130859375,
      "learning_rate": 8.94763946034463e-06,
      "loss": 142.2092,
      "step": 35800
    },
    {
      "epoch": 0.2962319559912313,
      "grad_norm": 966.3784790039062,
      "learning_rate": 8.946773740332781e-06,
      "loss": 157.32,
      "step": 35810
    },
    {
      "epoch": 0.29631467924060056,
      "grad_norm": 1265.8228759765625,
      "learning_rate": 8.945907706294262e-06,
      "loss": 87.6832,
      "step": 35820
    },
    {
      "epoch": 0.2963974024899698,
      "grad_norm": 434.8976135253906,
      "learning_rate": 8.945041358297973e-06,
      "loss": 115.3741,
      "step": 35830
    },
    {
      "epoch": 0.296480125739339,
      "grad_norm": 1037.9794921875,
      "learning_rate": 8.94417469641285e-06,
      "loss": 101.735,
      "step": 35840
    },
    {
      "epoch": 0.29656284898870827,
      "grad_norm": 1543.5999755859375,
      "learning_rate": 8.943307720707846e-06,
      "loss": 149.339,
      "step": 35850
    },
    {
      "epoch": 0.2966455722380775,
      "grad_norm": 851.0203247070312,
      "learning_rate": 8.942440431251947e-06,
      "loss": 126.8035,
      "step": 35860
    },
    {
      "epoch": 0.2967282954874467,
      "grad_norm": 1087.814697265625,
      "learning_rate": 8.941572828114154e-06,
      "loss": 154.3589,
      "step": 35870
    },
    {
      "epoch": 0.296811018736816,
      "grad_norm": 1085.237548828125,
      "learning_rate": 8.9407049113635e-06,
      "loss": 133.6983,
      "step": 35880
    },
    {
      "epoch": 0.29689374198618523,
      "grad_norm": 657.667724609375,
      "learning_rate": 8.939836681069042e-06,
      "loss": 77.88,
      "step": 35890
    },
    {
      "epoch": 0.29697646523555443,
      "grad_norm": 937.2635498046875,
      "learning_rate": 8.938968137299861e-06,
      "loss": 121.5767,
      "step": 35900
    },
    {
      "epoch": 0.2970591884849237,
      "grad_norm": 692.5191650390625,
      "learning_rate": 8.938099280125064e-06,
      "loss": 110.6443,
      "step": 35910
    },
    {
      "epoch": 0.29714191173429294,
      "grad_norm": 1191.0335693359375,
      "learning_rate": 8.937230109613778e-06,
      "loss": 125.6926,
      "step": 35920
    },
    {
      "epoch": 0.29722463498366214,
      "grad_norm": 600.857177734375,
      "learning_rate": 8.936360625835164e-06,
      "loss": 114.9589,
      "step": 35930
    },
    {
      "epoch": 0.2973073582330314,
      "grad_norm": 786.0075073242188,
      "learning_rate": 8.935490828858399e-06,
      "loss": 156.1116,
      "step": 35940
    },
    {
      "epoch": 0.29739008148240065,
      "grad_norm": 923.1871948242188,
      "learning_rate": 8.934620718752691e-06,
      "loss": 102.9856,
      "step": 35950
    },
    {
      "epoch": 0.29747280473176985,
      "grad_norm": 785.8792114257812,
      "learning_rate": 8.933750295587269e-06,
      "loss": 121.2862,
      "step": 35960
    },
    {
      "epoch": 0.2975555279811391,
      "grad_norm": 1262.131591796875,
      "learning_rate": 8.932879559431392e-06,
      "loss": 144.9996,
      "step": 35970
    },
    {
      "epoch": 0.29763825123050835,
      "grad_norm": 1094.29296875,
      "learning_rate": 8.932008510354336e-06,
      "loss": 99.4907,
      "step": 35980
    },
    {
      "epoch": 0.29772097447987755,
      "grad_norm": 1515.66748046875,
      "learning_rate": 8.931137148425407e-06,
      "loss": 117.0325,
      "step": 35990
    },
    {
      "epoch": 0.2978036977292468,
      "grad_norm": 702.0181274414062,
      "learning_rate": 8.930265473713939e-06,
      "loss": 89.8551,
      "step": 36000
    },
    {
      "epoch": 0.29788642097861606,
      "grad_norm": 924.1445922851562,
      "learning_rate": 8.929393486289283e-06,
      "loss": 91.7574,
      "step": 36010
    },
    {
      "epoch": 0.29796914422798526,
      "grad_norm": 886.2630615234375,
      "learning_rate": 8.928521186220822e-06,
      "loss": 134.8864,
      "step": 36020
    },
    {
      "epoch": 0.2980518674773545,
      "grad_norm": 571.9080200195312,
      "learning_rate": 8.92764857357796e-06,
      "loss": 166.6288,
      "step": 36030
    },
    {
      "epoch": 0.29813459072672377,
      "grad_norm": 1017.0745239257812,
      "learning_rate": 8.926775648430124e-06,
      "loss": 97.3446,
      "step": 36040
    },
    {
      "epoch": 0.29821731397609297,
      "grad_norm": 1384.15869140625,
      "learning_rate": 8.925902410846774e-06,
      "loss": 102.3454,
      "step": 36050
    },
    {
      "epoch": 0.2983000372254622,
      "grad_norm": 627.1011962890625,
      "learning_rate": 8.925028860897384e-06,
      "loss": 99.8053,
      "step": 36060
    },
    {
      "epoch": 0.2983827604748315,
      "grad_norm": 1164.1707763671875,
      "learning_rate": 8.924154998651461e-06,
      "loss": 150.2465,
      "step": 36070
    },
    {
      "epoch": 0.2984654837242007,
      "grad_norm": 806.7046508789062,
      "learning_rate": 8.923280824178538e-06,
      "loss": 127.2189,
      "step": 36080
    },
    {
      "epoch": 0.29854820697356993,
      "grad_norm": 936.9692993164062,
      "learning_rate": 8.922406337548162e-06,
      "loss": 106.9401,
      "step": 36090
    },
    {
      "epoch": 0.2986309302229392,
      "grad_norm": 861.8311157226562,
      "learning_rate": 8.921531538829917e-06,
      "loss": 92.577,
      "step": 36100
    },
    {
      "epoch": 0.2987136534723084,
      "grad_norm": 901.859375,
      "learning_rate": 8.920656428093403e-06,
      "loss": 83.8378,
      "step": 36110
    },
    {
      "epoch": 0.29879637672167764,
      "grad_norm": 1293.734375,
      "learning_rate": 8.919781005408251e-06,
      "loss": 114.6592,
      "step": 36120
    },
    {
      "epoch": 0.29887909997104684,
      "grad_norm": 697.5694580078125,
      "learning_rate": 8.918905270844113e-06,
      "loss": 141.8754,
      "step": 36130
    },
    {
      "epoch": 0.2989618232204161,
      "grad_norm": 822.4385986328125,
      "learning_rate": 8.918029224470671e-06,
      "loss": 101.8231,
      "step": 36140
    },
    {
      "epoch": 0.29904454646978534,
      "grad_norm": 886.5821533203125,
      "learning_rate": 8.917152866357621e-06,
      "loss": 97.08,
      "step": 36150
    },
    {
      "epoch": 0.29912726971915454,
      "grad_norm": 1727.11279296875,
      "learning_rate": 8.916276196574698e-06,
      "loss": 124.1994,
      "step": 36160
    },
    {
      "epoch": 0.2992099929685238,
      "grad_norm": 649.4053955078125,
      "learning_rate": 8.91539921519165e-06,
      "loss": 113.0979,
      "step": 36170
    },
    {
      "epoch": 0.29929271621789305,
      "grad_norm": 762.2130737304688,
      "learning_rate": 8.914521922278255e-06,
      "loss": 118.0666,
      "step": 36180
    },
    {
      "epoch": 0.29937543946726225,
      "grad_norm": 995.956787109375,
      "learning_rate": 8.913644317904317e-06,
      "loss": 125.7407,
      "step": 36190
    },
    {
      "epoch": 0.2994581627166315,
      "grad_norm": 1194.2001953125,
      "learning_rate": 8.912766402139662e-06,
      "loss": 124.1992,
      "step": 36200
    },
    {
      "epoch": 0.29954088596600076,
      "grad_norm": 1277.1484375,
      "learning_rate": 8.91188817505414e-06,
      "loss": 146.9492,
      "step": 36210
    },
    {
      "epoch": 0.29962360921536996,
      "grad_norm": 992.379638671875,
      "learning_rate": 8.91100963671763e-06,
      "loss": 107.3992,
      "step": 36220
    },
    {
      "epoch": 0.2997063324647392,
      "grad_norm": 820.5222778320312,
      "learning_rate": 8.910130787200032e-06,
      "loss": 93.3464,
      "step": 36230
    },
    {
      "epoch": 0.29978905571410847,
      "grad_norm": 2073.3251953125,
      "learning_rate": 8.909251626571273e-06,
      "loss": 101.3619,
      "step": 36240
    },
    {
      "epoch": 0.29987177896347766,
      "grad_norm": 767.80615234375,
      "learning_rate": 8.908372154901302e-06,
      "loss": 89.5982,
      "step": 36250
    },
    {
      "epoch": 0.2999545022128469,
      "grad_norm": 1339.1114501953125,
      "learning_rate": 8.907492372260096e-06,
      "loss": 118.8273,
      "step": 36260
    },
    {
      "epoch": 0.3000372254622162,
      "grad_norm": 1395.037353515625,
      "learning_rate": 8.906612278717657e-06,
      "loss": 114.5038,
      "step": 36270
    },
    {
      "epoch": 0.30011994871158537,
      "grad_norm": 1279.881591796875,
      "learning_rate": 8.905731874344005e-06,
      "loss": 110.8277,
      "step": 36280
    },
    {
      "epoch": 0.3002026719609546,
      "grad_norm": 1153.6810302734375,
      "learning_rate": 8.904851159209193e-06,
      "loss": 111.3379,
      "step": 36290
    },
    {
      "epoch": 0.3002853952103239,
      "grad_norm": 910.88623046875,
      "learning_rate": 8.903970133383297e-06,
      "loss": 83.3806,
      "step": 36300
    },
    {
      "epoch": 0.3003681184596931,
      "grad_norm": 993.2054443359375,
      "learning_rate": 8.903088796936414e-06,
      "loss": 117.134,
      "step": 36310
    },
    {
      "epoch": 0.30045084170906233,
      "grad_norm": 1444.7095947265625,
      "learning_rate": 8.902207149938667e-06,
      "loss": 118.84,
      "step": 36320
    },
    {
      "epoch": 0.3005335649584316,
      "grad_norm": 722.3483276367188,
      "learning_rate": 8.901325192460206e-06,
      "loss": 100.5878,
      "step": 36330
    },
    {
      "epoch": 0.3006162882078008,
      "grad_norm": 2218.97119140625,
      "learning_rate": 8.900442924571204e-06,
      "loss": 125.5503,
      "step": 36340
    },
    {
      "epoch": 0.30069901145717004,
      "grad_norm": 933.83984375,
      "learning_rate": 8.89956034634186e-06,
      "loss": 100.4451,
      "step": 36350
    },
    {
      "epoch": 0.3007817347065393,
      "grad_norm": 1003.083251953125,
      "learning_rate": 8.898677457842394e-06,
      "loss": 97.8874,
      "step": 36360
    },
    {
      "epoch": 0.3008644579559085,
      "grad_norm": 1543.9967041015625,
      "learning_rate": 8.897794259143057e-06,
      "loss": 144.2935,
      "step": 36370
    },
    {
      "epoch": 0.30094718120527775,
      "grad_norm": 1334.9381103515625,
      "learning_rate": 8.896910750314118e-06,
      "loss": 91.8307,
      "step": 36380
    },
    {
      "epoch": 0.301029904454647,
      "grad_norm": 1186.411865234375,
      "learning_rate": 8.896026931425876e-06,
      "loss": 131.1232,
      "step": 36390
    },
    {
      "epoch": 0.3011126277040162,
      "grad_norm": 1868.9034423828125,
      "learning_rate": 8.895142802548653e-06,
      "loss": 124.4849,
      "step": 36400
    },
    {
      "epoch": 0.30119535095338545,
      "grad_norm": 840.4161987304688,
      "learning_rate": 8.89425836375279e-06,
      "loss": 81.0873,
      "step": 36410
    },
    {
      "epoch": 0.3012780742027547,
      "grad_norm": 1163.7984619140625,
      "learning_rate": 8.893373615108663e-06,
      "loss": 136.477,
      "step": 36420
    },
    {
      "epoch": 0.3013607974521239,
      "grad_norm": 1086.72216796875,
      "learning_rate": 8.892488556686665e-06,
      "loss": 126.1113,
      "step": 36430
    },
    {
      "epoch": 0.30144352070149316,
      "grad_norm": 879.4309692382812,
      "learning_rate": 8.891603188557218e-06,
      "loss": 96.8926,
      "step": 36440
    },
    {
      "epoch": 0.3015262439508624,
      "grad_norm": 1197.5047607421875,
      "learning_rate": 8.890717510790763e-06,
      "loss": 110.1916,
      "step": 36450
    },
    {
      "epoch": 0.3016089672002316,
      "grad_norm": 349.7181701660156,
      "learning_rate": 8.889831523457773e-06,
      "loss": 154.8354,
      "step": 36460
    },
    {
      "epoch": 0.30169169044960087,
      "grad_norm": 1460.5762939453125,
      "learning_rate": 8.888945226628742e-06,
      "loss": 140.0803,
      "step": 36470
    },
    {
      "epoch": 0.3017744136989701,
      "grad_norm": 665.8645629882812,
      "learning_rate": 8.888058620374185e-06,
      "loss": 92.3529,
      "step": 36480
    },
    {
      "epoch": 0.3018571369483393,
      "grad_norm": 879.9156494140625,
      "learning_rate": 8.887171704764647e-06,
      "loss": 129.1003,
      "step": 36490
    },
    {
      "epoch": 0.3019398601977086,
      "grad_norm": 815.3402709960938,
      "learning_rate": 8.8862844798707e-06,
      "loss": 118.2584,
      "step": 36500
    },
    {
      "epoch": 0.3020225834470778,
      "grad_norm": 1314.0321044921875,
      "learning_rate": 8.885396945762928e-06,
      "loss": 126.3761,
      "step": 36510
    },
    {
      "epoch": 0.30210530669644703,
      "grad_norm": 858.8338012695312,
      "learning_rate": 8.884509102511956e-06,
      "loss": 98.5212,
      "step": 36520
    },
    {
      "epoch": 0.3021880299458163,
      "grad_norm": 1700.220458984375,
      "learning_rate": 8.883620950188422e-06,
      "loss": 106.6714,
      "step": 36530
    },
    {
      "epoch": 0.3022707531951855,
      "grad_norm": 696.0690307617188,
      "learning_rate": 8.882732488862988e-06,
      "loss": 118.2453,
      "step": 36540
    },
    {
      "epoch": 0.30235347644455474,
      "grad_norm": 793.1904907226562,
      "learning_rate": 8.881843718606353e-06,
      "loss": 125.0236,
      "step": 36550
    },
    {
      "epoch": 0.302436199693924,
      "grad_norm": 970.5441284179688,
      "learning_rate": 8.880954639489227e-06,
      "loss": 108.3671,
      "step": 36560
    },
    {
      "epoch": 0.3025189229432932,
      "grad_norm": 1777.4910888671875,
      "learning_rate": 8.880065251582354e-06,
      "loss": 160.0988,
      "step": 36570
    },
    {
      "epoch": 0.30260164619266244,
      "grad_norm": 1037.1177978515625,
      "learning_rate": 8.879175554956495e-06,
      "loss": 104.0029,
      "step": 36580
    },
    {
      "epoch": 0.3026843694420317,
      "grad_norm": 806.9476928710938,
      "learning_rate": 8.87828554968244e-06,
      "loss": 88.7504,
      "step": 36590
    },
    {
      "epoch": 0.3027670926914009,
      "grad_norm": 1243.7362060546875,
      "learning_rate": 8.877395235831002e-06,
      "loss": 111.0507,
      "step": 36600
    },
    {
      "epoch": 0.30284981594077015,
      "grad_norm": 1789.92529296875,
      "learning_rate": 8.876504613473019e-06,
      "loss": 154.0786,
      "step": 36610
    },
    {
      "epoch": 0.3029325391901394,
      "grad_norm": 718.7664184570312,
      "learning_rate": 8.875613682679356e-06,
      "loss": 108.9329,
      "step": 36620
    },
    {
      "epoch": 0.3030152624395086,
      "grad_norm": 748.9815063476562,
      "learning_rate": 8.874722443520898e-06,
      "loss": 123.7988,
      "step": 36630
    },
    {
      "epoch": 0.30309798568887786,
      "grad_norm": 606.9013671875,
      "learning_rate": 8.873830896068559e-06,
      "loss": 107.3505,
      "step": 36640
    },
    {
      "epoch": 0.3031807089382471,
      "grad_norm": 878.5547485351562,
      "learning_rate": 8.872939040393274e-06,
      "loss": 113.0779,
      "step": 36650
    },
    {
      "epoch": 0.3032634321876163,
      "grad_norm": 1300.362060546875,
      "learning_rate": 8.872046876566003e-06,
      "loss": 130.3682,
      "step": 36660
    },
    {
      "epoch": 0.30334615543698557,
      "grad_norm": 886.26025390625,
      "learning_rate": 8.871154404657734e-06,
      "loss": 106.1408,
      "step": 36670
    },
    {
      "epoch": 0.3034288786863548,
      "grad_norm": 1735.70947265625,
      "learning_rate": 8.870261624739474e-06,
      "loss": 120.6958,
      "step": 36680
    },
    {
      "epoch": 0.303511601935724,
      "grad_norm": 886.688232421875,
      "learning_rate": 8.869368536882258e-06,
      "loss": 102.5698,
      "step": 36690
    },
    {
      "epoch": 0.3035943251850933,
      "grad_norm": 632.3603515625,
      "learning_rate": 8.868475141157146e-06,
      "loss": 88.5606,
      "step": 36700
    },
    {
      "epoch": 0.3036770484344625,
      "grad_norm": 764.1319580078125,
      "learning_rate": 8.867581437635221e-06,
      "loss": 107.0108,
      "step": 36710
    },
    {
      "epoch": 0.3037597716838317,
      "grad_norm": 1273.7620849609375,
      "learning_rate": 8.866687426387592e-06,
      "loss": 159.5809,
      "step": 36720
    },
    {
      "epoch": 0.303842494933201,
      "grad_norm": 685.6810302734375,
      "learning_rate": 8.86579310748539e-06,
      "loss": 99.7657,
      "step": 36730
    },
    {
      "epoch": 0.30392521818257023,
      "grad_norm": 777.0698852539062,
      "learning_rate": 8.86489848099977e-06,
      "loss": 120.2746,
      "step": 36740
    },
    {
      "epoch": 0.30400794143193943,
      "grad_norm": 746.861572265625,
      "learning_rate": 8.864003547001916e-06,
      "loss": 106.8211,
      "step": 36750
    },
    {
      "epoch": 0.3040906646813087,
      "grad_norm": 853.1625366210938,
      "learning_rate": 8.863108305563035e-06,
      "loss": 91.7284,
      "step": 36760
    },
    {
      "epoch": 0.30417338793067794,
      "grad_norm": 1231.1922607421875,
      "learning_rate": 8.862212756754354e-06,
      "loss": 155.4766,
      "step": 36770
    },
    {
      "epoch": 0.30425611118004714,
      "grad_norm": 1215.257568359375,
      "learning_rate": 8.861316900647129e-06,
      "loss": 149.8827,
      "step": 36780
    },
    {
      "epoch": 0.3043388344294164,
      "grad_norm": 3567.228515625,
      "learning_rate": 8.860420737312638e-06,
      "loss": 121.5637,
      "step": 36790
    },
    {
      "epoch": 0.30442155767878565,
      "grad_norm": 694.6704711914062,
      "learning_rate": 8.859524266822188e-06,
      "loss": 101.825,
      "step": 36800
    },
    {
      "epoch": 0.30450428092815485,
      "grad_norm": 1621.055908203125,
      "learning_rate": 8.858627489247105e-06,
      "loss": 128.4847,
      "step": 36810
    },
    {
      "epoch": 0.3045870041775241,
      "grad_norm": 974.5038452148438,
      "learning_rate": 8.85773040465874e-06,
      "loss": 157.9853,
      "step": 36820
    },
    {
      "epoch": 0.30466972742689336,
      "grad_norm": 1332.6234130859375,
      "learning_rate": 8.856833013128472e-06,
      "loss": 107.964,
      "step": 36830
    },
    {
      "epoch": 0.30475245067626255,
      "grad_norm": 1167.7923583984375,
      "learning_rate": 8.855935314727702e-06,
      "loss": 109.8553,
      "step": 36840
    },
    {
      "epoch": 0.3048351739256318,
      "grad_norm": 704.8397216796875,
      "learning_rate": 8.855037309527854e-06,
      "loss": 115.8193,
      "step": 36850
    },
    {
      "epoch": 0.304917897175001,
      "grad_norm": 883.3186645507812,
      "learning_rate": 8.854138997600382e-06,
      "loss": 91.2245,
      "step": 36860
    },
    {
      "epoch": 0.30500062042437026,
      "grad_norm": 660.802001953125,
      "learning_rate": 8.853240379016757e-06,
      "loss": 88.3629,
      "step": 36870
    },
    {
      "epoch": 0.3050833436737395,
      "grad_norm": 1194.5704345703125,
      "learning_rate": 8.852341453848477e-06,
      "loss": 113.7338,
      "step": 36880
    },
    {
      "epoch": 0.3051660669231087,
      "grad_norm": 1196.6671142578125,
      "learning_rate": 8.851442222167068e-06,
      "loss": 96.392,
      "step": 36890
    },
    {
      "epoch": 0.30524879017247797,
      "grad_norm": 1096.315673828125,
      "learning_rate": 8.850542684044078e-06,
      "loss": 132.9981,
      "step": 36900
    },
    {
      "epoch": 0.3053315134218472,
      "grad_norm": 843.752685546875,
      "learning_rate": 8.849642839551079e-06,
      "loss": 120.9463,
      "step": 36910
    },
    {
      "epoch": 0.3054142366712164,
      "grad_norm": 1700.408203125,
      "learning_rate": 8.848742688759666e-06,
      "loss": 106.4173,
      "step": 36920
    },
    {
      "epoch": 0.3054969599205857,
      "grad_norm": 1714.5902099609375,
      "learning_rate": 8.847842231741462e-06,
      "loss": 119.7494,
      "step": 36930
    },
    {
      "epoch": 0.30557968316995493,
      "grad_norm": 998.3058471679688,
      "learning_rate": 8.846941468568108e-06,
      "loss": 104.204,
      "step": 36940
    },
    {
      "epoch": 0.30566240641932413,
      "grad_norm": 1132.1826171875,
      "learning_rate": 8.846040399311278e-06,
      "loss": 112.0986,
      "step": 36950
    },
    {
      "epoch": 0.3057451296686934,
      "grad_norm": 661.6875,
      "learning_rate": 8.845139024042664e-06,
      "loss": 81.744,
      "step": 36960
    },
    {
      "epoch": 0.30582785291806264,
      "grad_norm": 932.6510009765625,
      "learning_rate": 8.844237342833985e-06,
      "loss": 127.1373,
      "step": 36970
    },
    {
      "epoch": 0.30591057616743184,
      "grad_norm": 780.3368530273438,
      "learning_rate": 8.843335355756983e-06,
      "loss": 75.523,
      "step": 36980
    },
    {
      "epoch": 0.3059932994168011,
      "grad_norm": 754.1832275390625,
      "learning_rate": 8.842433062883427e-06,
      "loss": 89.4337,
      "step": 36990
    },
    {
      "epoch": 0.30607602266617034,
      "grad_norm": 764.4425659179688,
      "learning_rate": 8.841530464285105e-06,
      "loss": 129.0274,
      "step": 37000
    },
    {
      "epoch": 0.30615874591553954,
      "grad_norm": 861.7986450195312,
      "learning_rate": 8.840627560033833e-06,
      "loss": 82.3673,
      "step": 37010
    },
    {
      "epoch": 0.3062414691649088,
      "grad_norm": 988.38916015625,
      "learning_rate": 8.839724350201452e-06,
      "loss": 114.8146,
      "step": 37020
    },
    {
      "epoch": 0.30632419241427805,
      "grad_norm": 530.607421875,
      "learning_rate": 8.838820834859829e-06,
      "loss": 103.93,
      "step": 37030
    },
    {
      "epoch": 0.30640691566364725,
      "grad_norm": 941.2070922851562,
      "learning_rate": 8.837917014080849e-06,
      "loss": 89.9074,
      "step": 37040
    },
    {
      "epoch": 0.3064896389130165,
      "grad_norm": 688.3717651367188,
      "learning_rate": 8.837012887936426e-06,
      "loss": 111.2304,
      "step": 37050
    },
    {
      "epoch": 0.30657236216238576,
      "grad_norm": 1092.7115478515625,
      "learning_rate": 8.836108456498497e-06,
      "loss": 93.2177,
      "step": 37060
    },
    {
      "epoch": 0.30665508541175496,
      "grad_norm": 970.1156616210938,
      "learning_rate": 8.835203719839024e-06,
      "loss": 113.6382,
      "step": 37070
    },
    {
      "epoch": 0.3067378086611242,
      "grad_norm": 1119.5540771484375,
      "learning_rate": 8.834298678029988e-06,
      "loss": 120.2044,
      "step": 37080
    },
    {
      "epoch": 0.30682053191049347,
      "grad_norm": 840.0438842773438,
      "learning_rate": 8.833393331143409e-06,
      "loss": 103.9367,
      "step": 37090
    },
    {
      "epoch": 0.30690325515986266,
      "grad_norm": 1232.6182861328125,
      "learning_rate": 8.832487679251311e-06,
      "loss": 107.9942,
      "step": 37100
    },
    {
      "epoch": 0.3069859784092319,
      "grad_norm": 702.6771850585938,
      "learning_rate": 8.831581722425761e-06,
      "loss": 107.5534,
      "step": 37110
    },
    {
      "epoch": 0.3070687016586012,
      "grad_norm": 933.55615234375,
      "learning_rate": 8.830675460738835e-06,
      "loss": 92.3703,
      "step": 37120
    },
    {
      "epoch": 0.30715142490797037,
      "grad_norm": 1156.6966552734375,
      "learning_rate": 8.829768894262644e-06,
      "loss": 118.0975,
      "step": 37130
    },
    {
      "epoch": 0.3072341481573396,
      "grad_norm": 766.1238403320312,
      "learning_rate": 8.82886202306932e-06,
      "loss": 91.8914,
      "step": 37140
    },
    {
      "epoch": 0.3073168714067089,
      "grad_norm": 1076.5308837890625,
      "learning_rate": 8.827954847231016e-06,
      "loss": 115.0902,
      "step": 37150
    },
    {
      "epoch": 0.3073995946560781,
      "grad_norm": 867.0706176757812,
      "learning_rate": 8.82704736681991e-06,
      "loss": 128.4827,
      "step": 37160
    },
    {
      "epoch": 0.30748231790544733,
      "grad_norm": 963.2921752929688,
      "learning_rate": 8.826139581908211e-06,
      "loss": 112.9323,
      "step": 37170
    },
    {
      "epoch": 0.3075650411548166,
      "grad_norm": 735.8433227539062,
      "learning_rate": 8.825231492568146e-06,
      "loss": 114.1932,
      "step": 37180
    },
    {
      "epoch": 0.3076477644041858,
      "grad_norm": 1258.63818359375,
      "learning_rate": 8.824323098871966e-06,
      "loss": 136.7632,
      "step": 37190
    },
    {
      "epoch": 0.30773048765355504,
      "grad_norm": 1213.376220703125,
      "learning_rate": 8.823414400891948e-06,
      "loss": 140.9363,
      "step": 37200
    },
    {
      "epoch": 0.30781321090292424,
      "grad_norm": 816.0035400390625,
      "learning_rate": 8.822505398700395e-06,
      "loss": 97.0494,
      "step": 37210
    },
    {
      "epoch": 0.3078959341522935,
      "grad_norm": 908.9609375,
      "learning_rate": 8.821596092369627e-06,
      "loss": 112.1852,
      "step": 37220
    },
    {
      "epoch": 0.30797865740166275,
      "grad_norm": 1214.3897705078125,
      "learning_rate": 8.820686481971998e-06,
      "loss": 133.1782,
      "step": 37230
    },
    {
      "epoch": 0.30806138065103195,
      "grad_norm": 1155.935546875,
      "learning_rate": 8.81977656757988e-06,
      "loss": 91.5354,
      "step": 37240
    },
    {
      "epoch": 0.3081441039004012,
      "grad_norm": 1258.32275390625,
      "learning_rate": 8.81886634926567e-06,
      "loss": 108.436,
      "step": 37250
    },
    {
      "epoch": 0.30822682714977045,
      "grad_norm": 1130.4833984375,
      "learning_rate": 8.817955827101794e-06,
      "loss": 133.1508,
      "step": 37260
    },
    {
      "epoch": 0.30830955039913965,
      "grad_norm": 738.8419799804688,
      "learning_rate": 8.817045001160693e-06,
      "loss": 122.5803,
      "step": 37270
    },
    {
      "epoch": 0.3083922736485089,
      "grad_norm": 913.4530639648438,
      "learning_rate": 8.816133871514838e-06,
      "loss": 108.7282,
      "step": 37280
    },
    {
      "epoch": 0.30847499689787816,
      "grad_norm": 600.1754150390625,
      "learning_rate": 8.815222438236726e-06,
      "loss": 91.7117,
      "step": 37290
    },
    {
      "epoch": 0.30855772014724736,
      "grad_norm": 987.5612182617188,
      "learning_rate": 8.814310701398873e-06,
      "loss": 111.6003,
      "step": 37300
    },
    {
      "epoch": 0.3086404433966166,
      "grad_norm": 718.8853759765625,
      "learning_rate": 8.813398661073823e-06,
      "loss": 120.6641,
      "step": 37310
    },
    {
      "epoch": 0.30872316664598587,
      "grad_norm": 969.3789672851562,
      "learning_rate": 8.812486317334145e-06,
      "loss": 112.9521,
      "step": 37320
    },
    {
      "epoch": 0.30880588989535507,
      "grad_norm": 1069.70849609375,
      "learning_rate": 8.811573670252426e-06,
      "loss": 102.9678,
      "step": 37330
    },
    {
      "epoch": 0.3088886131447243,
      "grad_norm": 1160.0357666015625,
      "learning_rate": 8.810660719901283e-06,
      "loss": 142.8662,
      "step": 37340
    },
    {
      "epoch": 0.3089713363940936,
      "grad_norm": 923.1897583007812,
      "learning_rate": 8.809747466353356e-06,
      "loss": 134.5727,
      "step": 37350
    },
    {
      "epoch": 0.3090540596434628,
      "grad_norm": 1227.94580078125,
      "learning_rate": 8.808833909681305e-06,
      "loss": 144.0661,
      "step": 37360
    },
    {
      "epoch": 0.30913678289283203,
      "grad_norm": 636.2891235351562,
      "learning_rate": 8.80792004995782e-06,
      "loss": 120.0885,
      "step": 37370
    },
    {
      "epoch": 0.3092195061422013,
      "grad_norm": 1097.7366943359375,
      "learning_rate": 8.807005887255615e-06,
      "loss": 138.4748,
      "step": 37380
    },
    {
      "epoch": 0.3093022293915705,
      "grad_norm": 814.5217895507812,
      "learning_rate": 8.806091421647423e-06,
      "loss": 113.6995,
      "step": 37390
    },
    {
      "epoch": 0.30938495264093974,
      "grad_norm": 737.350341796875,
      "learning_rate": 8.805176653206004e-06,
      "loss": 116.3498,
      "step": 37400
    },
    {
      "epoch": 0.309467675890309,
      "grad_norm": 2945.936279296875,
      "learning_rate": 8.80426158200414e-06,
      "loss": 109.3581,
      "step": 37410
    },
    {
      "epoch": 0.3095503991396782,
      "grad_norm": 1349.6761474609375,
      "learning_rate": 8.803346208114643e-06,
      "loss": 117.0218,
      "step": 37420
    },
    {
      "epoch": 0.30963312238904744,
      "grad_norm": 744.024658203125,
      "learning_rate": 8.802430531610344e-06,
      "loss": 101.151,
      "step": 37430
    },
    {
      "epoch": 0.3097158456384167,
      "grad_norm": 1006.9130859375,
      "learning_rate": 8.801514552564097e-06,
      "loss": 87.0184,
      "step": 37440
    },
    {
      "epoch": 0.3097985688877859,
      "grad_norm": 1106.60693359375,
      "learning_rate": 8.800598271048784e-06,
      "loss": 159.6884,
      "step": 37450
    },
    {
      "epoch": 0.30988129213715515,
      "grad_norm": 1054.00244140625,
      "learning_rate": 8.799681687137309e-06,
      "loss": 97.845,
      "step": 37460
    },
    {
      "epoch": 0.3099640153865244,
      "grad_norm": 551.2701416015625,
      "learning_rate": 8.7987648009026e-06,
      "loss": 87.5821,
      "step": 37470
    },
    {
      "epoch": 0.3100467386358936,
      "grad_norm": 2352.765869140625,
      "learning_rate": 8.79784761241761e-06,
      "loss": 144.6846,
      "step": 37480
    },
    {
      "epoch": 0.31012946188526286,
      "grad_norm": 899.6213989257812,
      "learning_rate": 8.796930121755315e-06,
      "loss": 110.2097,
      "step": 37490
    },
    {
      "epoch": 0.3102121851346321,
      "grad_norm": 527.784912109375,
      "learning_rate": 8.796012328988716e-06,
      "loss": 98.5166,
      "step": 37500
    },
    {
      "epoch": 0.3102949083840013,
      "grad_norm": 1046.642822265625,
      "learning_rate": 8.795094234190837e-06,
      "loss": 94.5682,
      "step": 37510
    },
    {
      "epoch": 0.31037763163337057,
      "grad_norm": 1084.1600341796875,
      "learning_rate": 8.794175837434729e-06,
      "loss": 141.1946,
      "step": 37520
    },
    {
      "epoch": 0.3104603548827398,
      "grad_norm": 761.6773071289062,
      "learning_rate": 8.79325713879346e-06,
      "loss": 87.328,
      "step": 37530
    },
    {
      "epoch": 0.310543078132109,
      "grad_norm": 1374.055908203125,
      "learning_rate": 8.792338138340131e-06,
      "loss": 102.905,
      "step": 37540
    },
    {
      "epoch": 0.3106258013814783,
      "grad_norm": 935.9447631835938,
      "learning_rate": 8.791418836147858e-06,
      "loss": 146.4921,
      "step": 37550
    },
    {
      "epoch": 0.3107085246308475,
      "grad_norm": 1152.4400634765625,
      "learning_rate": 8.790499232289793e-06,
      "loss": 128.4351,
      "step": 37560
    },
    {
      "epoch": 0.3107912478802167,
      "grad_norm": 658.3630981445312,
      "learning_rate": 8.789579326839097e-06,
      "loss": 121.7294,
      "step": 37570
    },
    {
      "epoch": 0.310873971129586,
      "grad_norm": 830.0960083007812,
      "learning_rate": 8.788659119868966e-06,
      "loss": 133.9257,
      "step": 37580
    },
    {
      "epoch": 0.3109566943789552,
      "grad_norm": 866.0687255859375,
      "learning_rate": 8.787738611452616e-06,
      "loss": 116.4662,
      "step": 37590
    },
    {
      "epoch": 0.31103941762832443,
      "grad_norm": 1266.32177734375,
      "learning_rate": 8.78681780166329e-06,
      "loss": 120.4589,
      "step": 37600
    },
    {
      "epoch": 0.3111221408776937,
      "grad_norm": 876.0301513671875,
      "learning_rate": 8.785896690574248e-06,
      "loss": 103.7038,
      "step": 37610
    },
    {
      "epoch": 0.3112048641270629,
      "grad_norm": 822.327880859375,
      "learning_rate": 8.784975278258783e-06,
      "loss": 146.5088,
      "step": 37620
    },
    {
      "epoch": 0.31128758737643214,
      "grad_norm": 976.1931762695312,
      "learning_rate": 8.784053564790205e-06,
      "loss": 110.9248,
      "step": 37630
    },
    {
      "epoch": 0.3113703106258014,
      "grad_norm": 1754.4827880859375,
      "learning_rate": 8.783131550241853e-06,
      "loss": 132.1888,
      "step": 37640
    },
    {
      "epoch": 0.3114530338751706,
      "grad_norm": 889.1416015625,
      "learning_rate": 8.782209234687083e-06,
      "loss": 98.5607,
      "step": 37650
    },
    {
      "epoch": 0.31153575712453985,
      "grad_norm": 578.06689453125,
      "learning_rate": 8.781286618199285e-06,
      "loss": 93.0681,
      "step": 37660
    },
    {
      "epoch": 0.3116184803739091,
      "grad_norm": 625.0341796875,
      "learning_rate": 8.780363700851863e-06,
      "loss": 84.4234,
      "step": 37670
    },
    {
      "epoch": 0.3117012036232783,
      "grad_norm": 1180.2471923828125,
      "learning_rate": 8.779440482718251e-06,
      "loss": 126.0896,
      "step": 37680
    },
    {
      "epoch": 0.31178392687264755,
      "grad_norm": 1036.988525390625,
      "learning_rate": 8.778516963871904e-06,
      "loss": 109.445,
      "step": 37690
    },
    {
      "epoch": 0.3118666501220168,
      "grad_norm": 771.7061157226562,
      "learning_rate": 8.777593144386305e-06,
      "loss": 106.4233,
      "step": 37700
    },
    {
      "epoch": 0.311949373371386,
      "grad_norm": 853.5861206054688,
      "learning_rate": 8.776669024334955e-06,
      "loss": 149.2146,
      "step": 37710
    },
    {
      "epoch": 0.31203209662075526,
      "grad_norm": 1038.280029296875,
      "learning_rate": 8.775744603791385e-06,
      "loss": 87.6942,
      "step": 37720
    },
    {
      "epoch": 0.3121148198701245,
      "grad_norm": 785.39892578125,
      "learning_rate": 8.774819882829144e-06,
      "loss": 101.3138,
      "step": 37730
    },
    {
      "epoch": 0.3121975431194937,
      "grad_norm": 810.301513671875,
      "learning_rate": 8.77389486152181e-06,
      "loss": 104.7605,
      "step": 37740
    },
    {
      "epoch": 0.31228026636886297,
      "grad_norm": 907.918212890625,
      "learning_rate": 8.772969539942981e-06,
      "loss": 93.2778,
      "step": 37750
    },
    {
      "epoch": 0.3123629896182322,
      "grad_norm": 1630.78955078125,
      "learning_rate": 8.772043918166282e-06,
      "loss": 118.6197,
      "step": 37760
    },
    {
      "epoch": 0.3124457128676014,
      "grad_norm": 792.9736328125,
      "learning_rate": 8.771117996265358e-06,
      "loss": 110.336,
      "step": 37770
    },
    {
      "epoch": 0.3125284361169707,
      "grad_norm": 515.8519287109375,
      "learning_rate": 8.770191774313883e-06,
      "loss": 98.6496,
      "step": 37780
    },
    {
      "epoch": 0.31261115936633993,
      "grad_norm": 889.776611328125,
      "learning_rate": 8.769265252385552e-06,
      "loss": 165.6621,
      "step": 37790
    },
    {
      "epoch": 0.31269388261570913,
      "grad_norm": 780.7835693359375,
      "learning_rate": 8.768338430554083e-06,
      "loss": 91.5842,
      "step": 37800
    },
    {
      "epoch": 0.3127766058650784,
      "grad_norm": 1138.0804443359375,
      "learning_rate": 8.76741130889322e-06,
      "loss": 100.7134,
      "step": 37810
    },
    {
      "epoch": 0.31285932911444764,
      "grad_norm": 942.5914916992188,
      "learning_rate": 8.766483887476727e-06,
      "loss": 105.6115,
      "step": 37820
    },
    {
      "epoch": 0.31294205236381684,
      "grad_norm": 618.5468139648438,
      "learning_rate": 8.7655561663784e-06,
      "loss": 75.4478,
      "step": 37830
    },
    {
      "epoch": 0.3130247756131861,
      "grad_norm": 1693.069580078125,
      "learning_rate": 8.764628145672048e-06,
      "loss": 127.7167,
      "step": 37840
    },
    {
      "epoch": 0.31310749886255534,
      "grad_norm": 1026.7965087890625,
      "learning_rate": 8.763699825431513e-06,
      "loss": 138.911,
      "step": 37850
    },
    {
      "epoch": 0.31319022211192454,
      "grad_norm": 1648.09765625,
      "learning_rate": 8.762771205730656e-06,
      "loss": 117.1393,
      "step": 37860
    },
    {
      "epoch": 0.3132729453612938,
      "grad_norm": 933.2096557617188,
      "learning_rate": 8.761842286643362e-06,
      "loss": 96.2922,
      "step": 37870
    },
    {
      "epoch": 0.31335566861066305,
      "grad_norm": 802.694580078125,
      "learning_rate": 8.760913068243542e-06,
      "loss": 144.4842,
      "step": 37880
    },
    {
      "epoch": 0.31343839186003225,
      "grad_norm": 1243.0850830078125,
      "learning_rate": 8.759983550605132e-06,
      "loss": 128.3055,
      "step": 37890
    },
    {
      "epoch": 0.3135211151094015,
      "grad_norm": 1257.254638671875,
      "learning_rate": 8.759053733802083e-06,
      "loss": 100.2831,
      "step": 37900
    },
    {
      "epoch": 0.31360383835877076,
      "grad_norm": 934.783935546875,
      "learning_rate": 8.758123617908383e-06,
      "loss": 100.1143,
      "step": 37910
    },
    {
      "epoch": 0.31368656160813996,
      "grad_norm": 989.8101196289062,
      "learning_rate": 8.757193202998033e-06,
      "loss": 127.9963,
      "step": 37920
    },
    {
      "epoch": 0.3137692848575092,
      "grad_norm": 1011.3324584960938,
      "learning_rate": 8.756262489145061e-06,
      "loss": 112.5696,
      "step": 37930
    },
    {
      "epoch": 0.3138520081068784,
      "grad_norm": 925.4760131835938,
      "learning_rate": 8.755331476423526e-06,
      "loss": 79.9976,
      "step": 37940
    },
    {
      "epoch": 0.31393473135624766,
      "grad_norm": 932.2153930664062,
      "learning_rate": 8.754400164907496e-06,
      "loss": 118.1142,
      "step": 37950
    },
    {
      "epoch": 0.3140174546056169,
      "grad_norm": 869.6834106445312,
      "learning_rate": 8.753468554671078e-06,
      "loss": 122.9429,
      "step": 37960
    },
    {
      "epoch": 0.3141001778549861,
      "grad_norm": 1097.6431884765625,
      "learning_rate": 8.752536645788391e-06,
      "loss": 116.1235,
      "step": 37970
    },
    {
      "epoch": 0.31418290110435537,
      "grad_norm": 943.837158203125,
      "learning_rate": 8.751604438333587e-06,
      "loss": 120.1827,
      "step": 37980
    },
    {
      "epoch": 0.3142656243537246,
      "grad_norm": 965.2254028320312,
      "learning_rate": 8.750671932380834e-06,
      "loss": 111.8385,
      "step": 37990
    },
    {
      "epoch": 0.3143483476030938,
      "grad_norm": 2054.935546875,
      "learning_rate": 8.749739128004329e-06,
      "loss": 105.5353,
      "step": 38000
    },
    {
      "epoch": 0.3144310708524631,
      "grad_norm": 885.7263793945312,
      "learning_rate": 8.748806025278292e-06,
      "loss": 113.4429,
      "step": 38010
    },
    {
      "epoch": 0.31451379410183233,
      "grad_norm": 1407.1341552734375,
      "learning_rate": 8.747872624276963e-06,
      "loss": 99.107,
      "step": 38020
    },
    {
      "epoch": 0.31459651735120153,
      "grad_norm": 1092.1849365234375,
      "learning_rate": 8.746938925074609e-06,
      "loss": 130.1728,
      "step": 38030
    },
    {
      "epoch": 0.3146792406005708,
      "grad_norm": 977.42578125,
      "learning_rate": 8.746004927745522e-06,
      "loss": 116.9955,
      "step": 38040
    },
    {
      "epoch": 0.31476196384994004,
      "grad_norm": 813.2716064453125,
      "learning_rate": 8.745070632364014e-06,
      "loss": 103.2874,
      "step": 38050
    },
    {
      "epoch": 0.31484468709930924,
      "grad_norm": 823.4459838867188,
      "learning_rate": 8.744136039004422e-06,
      "loss": 122.7185,
      "step": 38060
    },
    {
      "epoch": 0.3149274103486785,
      "grad_norm": 675.5857543945312,
      "learning_rate": 8.743201147741112e-06,
      "loss": 117.528,
      "step": 38070
    },
    {
      "epoch": 0.31501013359804775,
      "grad_norm": 1719.105224609375,
      "learning_rate": 8.742265958648464e-06,
      "loss": 110.0581,
      "step": 38080
    },
    {
      "epoch": 0.31509285684741695,
      "grad_norm": 1066.7659912109375,
      "learning_rate": 8.741330471800888e-06,
      "loss": 89.2473,
      "step": 38090
    },
    {
      "epoch": 0.3151755800967862,
      "grad_norm": 776.9163818359375,
      "learning_rate": 8.740394687272817e-06,
      "loss": 124.1178,
      "step": 38100
    },
    {
      "epoch": 0.31525830334615546,
      "grad_norm": 760.9815063476562,
      "learning_rate": 8.739458605138706e-06,
      "loss": 119.1256,
      "step": 38110
    },
    {
      "epoch": 0.31534102659552465,
      "grad_norm": 974.4657592773438,
      "learning_rate": 8.738522225473036e-06,
      "loss": 105.6252,
      "step": 38120
    },
    {
      "epoch": 0.3154237498448939,
      "grad_norm": 634.0014038085938,
      "learning_rate": 8.737585548350312e-06,
      "loss": 119.6853,
      "step": 38130
    },
    {
      "epoch": 0.31550647309426316,
      "grad_norm": 765.5140380859375,
      "learning_rate": 8.736648573845057e-06,
      "loss": 99.3297,
      "step": 38140
    },
    {
      "epoch": 0.31558919634363236,
      "grad_norm": 1316.4825439453125,
      "learning_rate": 8.735711302031824e-06,
      "loss": 123.088,
      "step": 38150
    },
    {
      "epoch": 0.3156719195930016,
      "grad_norm": 1433.5438232421875,
      "learning_rate": 8.734773732985186e-06,
      "loss": 116.7357,
      "step": 38160
    },
    {
      "epoch": 0.31575464284237087,
      "grad_norm": 826.6362915039062,
      "learning_rate": 8.733835866779745e-06,
      "loss": 93.5203,
      "step": 38170
    },
    {
      "epoch": 0.31583736609174007,
      "grad_norm": 1368.983642578125,
      "learning_rate": 8.73289770349012e-06,
      "loss": 90.895,
      "step": 38180
    },
    {
      "epoch": 0.3159200893411093,
      "grad_norm": 764.3739013671875,
      "learning_rate": 8.731959243190955e-06,
      "loss": 115.1358,
      "step": 38190
    },
    {
      "epoch": 0.3160028125904786,
      "grad_norm": 1166.0986328125,
      "learning_rate": 8.73102048595692e-06,
      "loss": 110.7284,
      "step": 38200
    },
    {
      "epoch": 0.3160855358398478,
      "grad_norm": 900.78759765625,
      "learning_rate": 8.730081431862709e-06,
      "loss": 114.7286,
      "step": 38210
    },
    {
      "epoch": 0.31616825908921703,
      "grad_norm": 694.7711791992188,
      "learning_rate": 8.729142080983037e-06,
      "loss": 99.7621,
      "step": 38220
    },
    {
      "epoch": 0.3162509823385863,
      "grad_norm": 1478.10546875,
      "learning_rate": 8.728202433392645e-06,
      "loss": 103.1368,
      "step": 38230
    },
    {
      "epoch": 0.3163337055879555,
      "grad_norm": 739.2174682617188,
      "learning_rate": 8.727262489166295e-06,
      "loss": 91.7107,
      "step": 38240
    },
    {
      "epoch": 0.31641642883732474,
      "grad_norm": 1041.38525390625,
      "learning_rate": 8.726322248378775e-06,
      "loss": 133.2948,
      "step": 38250
    },
    {
      "epoch": 0.316499152086694,
      "grad_norm": 709.4502563476562,
      "learning_rate": 8.725381711104894e-06,
      "loss": 134.3007,
      "step": 38260
    },
    {
      "epoch": 0.3165818753360632,
      "grad_norm": 1081.0482177734375,
      "learning_rate": 8.724440877419487e-06,
      "loss": 107.7189,
      "step": 38270
    },
    {
      "epoch": 0.31666459858543244,
      "grad_norm": 1248.0484619140625,
      "learning_rate": 8.723499747397415e-06,
      "loss": 105.8039,
      "step": 38280
    },
    {
      "epoch": 0.3167473218348017,
      "grad_norm": 1431.8619384765625,
      "learning_rate": 8.722558321113555e-06,
      "loss": 108.0174,
      "step": 38290
    },
    {
      "epoch": 0.3168300450841709,
      "grad_norm": 972.2454223632812,
      "learning_rate": 8.721616598642812e-06,
      "loss": 124.3465,
      "step": 38300
    },
    {
      "epoch": 0.31691276833354015,
      "grad_norm": 1106.1824951171875,
      "learning_rate": 8.720674580060117e-06,
      "loss": 92.7966,
      "step": 38310
    },
    {
      "epoch": 0.31699549158290935,
      "grad_norm": 1259.550048828125,
      "learning_rate": 8.719732265440423e-06,
      "loss": 91.808,
      "step": 38320
    },
    {
      "epoch": 0.3170782148322786,
      "grad_norm": 1003.4591064453125,
      "learning_rate": 8.718789654858702e-06,
      "loss": 97.9086,
      "step": 38330
    },
    {
      "epoch": 0.31716093808164786,
      "grad_norm": 1087.510498046875,
      "learning_rate": 8.717846748389956e-06,
      "loss": 190.7288,
      "step": 38340
    },
    {
      "epoch": 0.31724366133101706,
      "grad_norm": 1307.7303466796875,
      "learning_rate": 8.716903546109208e-06,
      "loss": 100.8898,
      "step": 38350
    },
    {
      "epoch": 0.3173263845803863,
      "grad_norm": 389.17926025390625,
      "learning_rate": 8.715960048091502e-06,
      "loss": 105.3628,
      "step": 38360
    },
    {
      "epoch": 0.31740910782975557,
      "grad_norm": 634.9244995117188,
      "learning_rate": 8.715016254411908e-06,
      "loss": 93.6207,
      "step": 38370
    },
    {
      "epoch": 0.31749183107912476,
      "grad_norm": 1184.2578125,
      "learning_rate": 8.714072165145521e-06,
      "loss": 135.4022,
      "step": 38380
    },
    {
      "epoch": 0.317574554328494,
      "grad_norm": 619.5433349609375,
      "learning_rate": 8.713127780367458e-06,
      "loss": 82.4095,
      "step": 38390
    },
    {
      "epoch": 0.3176572775778633,
      "grad_norm": 1421.2760009765625,
      "learning_rate": 8.712183100152858e-06,
      "loss": 104.6683,
      "step": 38400
    },
    {
      "epoch": 0.31774000082723247,
      "grad_norm": 569.2383422851562,
      "learning_rate": 8.711238124576884e-06,
      "loss": 100.9128,
      "step": 38410
    },
    {
      "epoch": 0.3178227240766017,
      "grad_norm": 486.7202453613281,
      "learning_rate": 8.710292853714726e-06,
      "loss": 102.4042,
      "step": 38420
    },
    {
      "epoch": 0.317905447325971,
      "grad_norm": 1279.0054931640625,
      "learning_rate": 8.709347287641593e-06,
      "loss": 107.4948,
      "step": 38430
    },
    {
      "epoch": 0.3179881705753402,
      "grad_norm": 777.2363891601562,
      "learning_rate": 8.70840142643272e-06,
      "loss": 126.2411,
      "step": 38440
    },
    {
      "epoch": 0.31807089382470943,
      "grad_norm": 680.6844482421875,
      "learning_rate": 8.707455270163365e-06,
      "loss": 98.6142,
      "step": 38450
    },
    {
      "epoch": 0.3181536170740787,
      "grad_norm": 1288.6944580078125,
      "learning_rate": 8.70650881890881e-06,
      "loss": 125.0557,
      "step": 38460
    },
    {
      "epoch": 0.3182363403234479,
      "grad_norm": 760.6031494140625,
      "learning_rate": 8.705562072744358e-06,
      "loss": 138.251,
      "step": 38470
    },
    {
      "epoch": 0.31831906357281714,
      "grad_norm": 883.0706176757812,
      "learning_rate": 8.704615031745337e-06,
      "loss": 111.4153,
      "step": 38480
    },
    {
      "epoch": 0.3184017868221864,
      "grad_norm": 1131.5177001953125,
      "learning_rate": 8.703667695987102e-06,
      "loss": 113.8998,
      "step": 38490
    },
    {
      "epoch": 0.3184845100715556,
      "grad_norm": 973.4132690429688,
      "learning_rate": 8.702720065545024e-06,
      "loss": 106.239,
      "step": 38500
    },
    {
      "epoch": 0.31856723332092485,
      "grad_norm": 1401.8857421875,
      "learning_rate": 8.701772140494504e-06,
      "loss": 131.2176,
      "step": 38510
    },
    {
      "epoch": 0.3186499565702941,
      "grad_norm": 735.4816284179688,
      "learning_rate": 8.700823920910964e-06,
      "loss": 124.5568,
      "step": 38520
    },
    {
      "epoch": 0.3187326798196633,
      "grad_norm": 1129.6258544921875,
      "learning_rate": 8.699875406869848e-06,
      "loss": 103.7197,
      "step": 38530
    },
    {
      "epoch": 0.31881540306903255,
      "grad_norm": 802.9367065429688,
      "learning_rate": 8.69892659844663e-06,
      "loss": 129.3652,
      "step": 38540
    },
    {
      "epoch": 0.3188981263184018,
      "grad_norm": 1019.4291381835938,
      "learning_rate": 8.697977495716794e-06,
      "loss": 113.1963,
      "step": 38550
    },
    {
      "epoch": 0.318980849567771,
      "grad_norm": 1141.768798828125,
      "learning_rate": 8.697028098755863e-06,
      "loss": 75.0446,
      "step": 38560
    },
    {
      "epoch": 0.31906357281714026,
      "grad_norm": 561.0872192382812,
      "learning_rate": 8.69607840763937e-06,
      "loss": 107.2292,
      "step": 38570
    },
    {
      "epoch": 0.3191462960665095,
      "grad_norm": 567.6842651367188,
      "learning_rate": 8.695128422442882e-06,
      "loss": 105.8062,
      "step": 38580
    },
    {
      "epoch": 0.3192290193158787,
      "grad_norm": 1812.7880859375,
      "learning_rate": 8.694178143241984e-06,
      "loss": 116.8599,
      "step": 38590
    },
    {
      "epoch": 0.31931174256524797,
      "grad_norm": 821.3114013671875,
      "learning_rate": 8.693227570112285e-06,
      "loss": 113.9192,
      "step": 38600
    },
    {
      "epoch": 0.3193944658146172,
      "grad_norm": 1383.909423828125,
      "learning_rate": 8.692276703129421e-06,
      "loss": 123.0948,
      "step": 38610
    },
    {
      "epoch": 0.3194771890639864,
      "grad_norm": 559.3286743164062,
      "learning_rate": 8.691325542369041e-06,
      "loss": 81.4486,
      "step": 38620
    },
    {
      "epoch": 0.3195599123133557,
      "grad_norm": 1106.412109375,
      "learning_rate": 8.69037408790683e-06,
      "loss": 119.6945,
      "step": 38630
    },
    {
      "epoch": 0.31964263556272493,
      "grad_norm": 1364.677490234375,
      "learning_rate": 8.689422339818489e-06,
      "loss": 140.9282,
      "step": 38640
    },
    {
      "epoch": 0.31972535881209413,
      "grad_norm": 0.0,
      "learning_rate": 8.688470298179746e-06,
      "loss": 140.5661,
      "step": 38650
    },
    {
      "epoch": 0.3198080820614634,
      "grad_norm": 859.6204833984375,
      "learning_rate": 8.687517963066347e-06,
      "loss": 110.4718,
      "step": 38660
    },
    {
      "epoch": 0.3198908053108326,
      "grad_norm": 1258.269775390625,
      "learning_rate": 8.686565334554069e-06,
      "loss": 126.0004,
      "step": 38670
    },
    {
      "epoch": 0.31997352856020184,
      "grad_norm": 1039.6004638671875,
      "learning_rate": 8.685612412718704e-06,
      "loss": 119.8658,
      "step": 38680
    },
    {
      "epoch": 0.3200562518095711,
      "grad_norm": 923.8244018554688,
      "learning_rate": 8.684659197636076e-06,
      "loss": 124.2017,
      "step": 38690
    },
    {
      "epoch": 0.3201389750589403,
      "grad_norm": 1117.4451904296875,
      "learning_rate": 8.683705689382025e-06,
      "loss": 107.7295,
      "step": 38700
    },
    {
      "epoch": 0.32022169830830954,
      "grad_norm": 1363.1929931640625,
      "learning_rate": 8.682751888032419e-06,
      "loss": 99.3945,
      "step": 38710
    },
    {
      "epoch": 0.3203044215576788,
      "grad_norm": 701.418212890625,
      "learning_rate": 8.681797793663147e-06,
      "loss": 120.6914,
      "step": 38720
    },
    {
      "epoch": 0.320387144807048,
      "grad_norm": 1962.092041015625,
      "learning_rate": 8.680843406350122e-06,
      "loss": 105.0907,
      "step": 38730
    },
    {
      "epoch": 0.32046986805641725,
      "grad_norm": 1574.8798828125,
      "learning_rate": 8.679888726169277e-06,
      "loss": 123.2075,
      "step": 38740
    },
    {
      "epoch": 0.3205525913057865,
      "grad_norm": 979.2647094726562,
      "learning_rate": 8.678933753196577e-06,
      "loss": 117.9523,
      "step": 38750
    },
    {
      "epoch": 0.3206353145551557,
      "grad_norm": 961.4496459960938,
      "learning_rate": 8.677978487508002e-06,
      "loss": 130.495,
      "step": 38760
    },
    {
      "epoch": 0.32071803780452496,
      "grad_norm": 1239.4903564453125,
      "learning_rate": 8.677022929179558e-06,
      "loss": 116.15,
      "step": 38770
    },
    {
      "epoch": 0.3208007610538942,
      "grad_norm": 936.8592529296875,
      "learning_rate": 8.676067078287276e-06,
      "loss": 102.5058,
      "step": 38780
    },
    {
      "epoch": 0.3208834843032634,
      "grad_norm": 1102.019775390625,
      "learning_rate": 8.675110934907206e-06,
      "loss": 105.1739,
      "step": 38790
    },
    {
      "epoch": 0.32096620755263267,
      "grad_norm": 1137.979736328125,
      "learning_rate": 8.674154499115426e-06,
      "loss": 103.8995,
      "step": 38800
    },
    {
      "epoch": 0.3210489308020019,
      "grad_norm": 983.4441528320312,
      "learning_rate": 8.673197770988034e-06,
      "loss": 100.5983,
      "step": 38810
    },
    {
      "epoch": 0.3211316540513711,
      "grad_norm": 601.077392578125,
      "learning_rate": 8.672240750601152e-06,
      "loss": 100.5274,
      "step": 38820
    },
    {
      "epoch": 0.3212143773007404,
      "grad_norm": 897.4487915039062,
      "learning_rate": 8.67128343803093e-06,
      "loss": 99.9841,
      "step": 38830
    },
    {
      "epoch": 0.3212971005501096,
      "grad_norm": 1025.739013671875,
      "learning_rate": 8.670325833353532e-06,
      "loss": 89.5816,
      "step": 38840
    },
    {
      "epoch": 0.3213798237994788,
      "grad_norm": 677.2406005859375,
      "learning_rate": 8.669367936645152e-06,
      "loss": 105.6764,
      "step": 38850
    },
    {
      "epoch": 0.3214625470488481,
      "grad_norm": 912.070068359375,
      "learning_rate": 8.668409747982005e-06,
      "loss": 129.2276,
      "step": 38860
    },
    {
      "epoch": 0.32154527029821733,
      "grad_norm": 909.7315673828125,
      "learning_rate": 8.667451267440332e-06,
      "loss": 98.6507,
      "step": 38870
    },
    {
      "epoch": 0.32162799354758653,
      "grad_norm": 934.35400390625,
      "learning_rate": 8.666492495096391e-06,
      "loss": 121.8479,
      "step": 38880
    },
    {
      "epoch": 0.3217107167969558,
      "grad_norm": 1550.7830810546875,
      "learning_rate": 8.66553343102647e-06,
      "loss": 124.7869,
      "step": 38890
    },
    {
      "epoch": 0.32179344004632504,
      "grad_norm": 1913.998291015625,
      "learning_rate": 8.664574075306876e-06,
      "loss": 109.4713,
      "step": 38900
    },
    {
      "epoch": 0.32187616329569424,
      "grad_norm": 665.229736328125,
      "learning_rate": 8.66361442801394e-06,
      "loss": 94.5389,
      "step": 38910
    },
    {
      "epoch": 0.3219588865450635,
      "grad_norm": 762.2039794921875,
      "learning_rate": 8.662654489224018e-06,
      "loss": 94.5307,
      "step": 38920
    },
    {
      "epoch": 0.32204160979443275,
      "grad_norm": 1010.405029296875,
      "learning_rate": 8.661694259013489e-06,
      "loss": 143.722,
      "step": 38930
    },
    {
      "epoch": 0.32212433304380195,
      "grad_norm": 808.5379638671875,
      "learning_rate": 8.660733737458751e-06,
      "loss": 134.8724,
      "step": 38940
    },
    {
      "epoch": 0.3222070562931712,
      "grad_norm": 964.7113037109375,
      "learning_rate": 8.659772924636232e-06,
      "loss": 122.8288,
      "step": 38950
    },
    {
      "epoch": 0.32228977954254046,
      "grad_norm": 908.0009765625,
      "learning_rate": 8.658811820622376e-06,
      "loss": 90.802,
      "step": 38960
    },
    {
      "epoch": 0.32237250279190965,
      "grad_norm": 1564.2470703125,
      "learning_rate": 8.657850425493656e-06,
      "loss": 129.7668,
      "step": 38970
    },
    {
      "epoch": 0.3224552260412789,
      "grad_norm": 1255.0084228515625,
      "learning_rate": 8.656888739326564e-06,
      "loss": 96.6529,
      "step": 38980
    },
    {
      "epoch": 0.32253794929064816,
      "grad_norm": 960.5479125976562,
      "learning_rate": 8.65592676219762e-06,
      "loss": 103.2335,
      "step": 38990
    },
    {
      "epoch": 0.32262067254001736,
      "grad_norm": 815.9246826171875,
      "learning_rate": 8.65496449418336e-06,
      "loss": 94.21,
      "step": 39000
    },
    {
      "epoch": 0.3227033957893866,
      "grad_norm": 942.2803344726562,
      "learning_rate": 8.654001935360349e-06,
      "loss": 108.4447,
      "step": 39010
    },
    {
      "epoch": 0.32278611903875587,
      "grad_norm": 1212.7076416015625,
      "learning_rate": 8.653039085805174e-06,
      "loss": 97.4576,
      "step": 39020
    },
    {
      "epoch": 0.32286884228812507,
      "grad_norm": 1259.3065185546875,
      "learning_rate": 8.652075945594444e-06,
      "loss": 96.3901,
      "step": 39030
    },
    {
      "epoch": 0.3229515655374943,
      "grad_norm": 673.8052368164062,
      "learning_rate": 8.651112514804793e-06,
      "loss": 94.6694,
      "step": 39040
    },
    {
      "epoch": 0.3230342887868635,
      "grad_norm": 1086.0980224609375,
      "learning_rate": 8.650148793512874e-06,
      "loss": 161.4135,
      "step": 39050
    },
    {
      "epoch": 0.3231170120362328,
      "grad_norm": 1190.9241943359375,
      "learning_rate": 8.649184781795367e-06,
      "loss": 122.2091,
      "step": 39060
    },
    {
      "epoch": 0.32319973528560203,
      "grad_norm": 750.9871215820312,
      "learning_rate": 8.648220479728976e-06,
      "loss": 129.1647,
      "step": 39070
    },
    {
      "epoch": 0.32328245853497123,
      "grad_norm": 675.1224365234375,
      "learning_rate": 8.647255887390425e-06,
      "loss": 87.7561,
      "step": 39080
    },
    {
      "epoch": 0.3233651817843405,
      "grad_norm": 960.1796875,
      "learning_rate": 8.64629100485646e-06,
      "loss": 107.4958,
      "step": 39090
    },
    {
      "epoch": 0.32344790503370974,
      "grad_norm": 1386.735595703125,
      "learning_rate": 8.645325832203855e-06,
      "loss": 135.6421,
      "step": 39100
    },
    {
      "epoch": 0.32353062828307894,
      "grad_norm": 589.93896484375,
      "learning_rate": 8.644360369509403e-06,
      "loss": 102.7022,
      "step": 39110
    },
    {
      "epoch": 0.3236133515324482,
      "grad_norm": 1394.318603515625,
      "learning_rate": 8.64339461684992e-06,
      "loss": 109.0257,
      "step": 39120
    },
    {
      "epoch": 0.32369607478181744,
      "grad_norm": 1033.558349609375,
      "learning_rate": 8.64242857430225e-06,
      "loss": 120.0752,
      "step": 39130
    },
    {
      "epoch": 0.32377879803118664,
      "grad_norm": 877.220947265625,
      "learning_rate": 8.641462241943255e-06,
      "loss": 149.2554,
      "step": 39140
    },
    {
      "epoch": 0.3238615212805559,
      "grad_norm": 1799.2706298828125,
      "learning_rate": 8.640495619849821e-06,
      "loss": 106.8699,
      "step": 39150
    },
    {
      "epoch": 0.32394424452992515,
      "grad_norm": 354.8163146972656,
      "learning_rate": 8.639528708098858e-06,
      "loss": 138.8774,
      "step": 39160
    },
    {
      "epoch": 0.32402696777929435,
      "grad_norm": 734.3604125976562,
      "learning_rate": 8.6385615067673e-06,
      "loss": 113.2349,
      "step": 39170
    },
    {
      "epoch": 0.3241096910286636,
      "grad_norm": 1038.3038330078125,
      "learning_rate": 8.6375940159321e-06,
      "loss": 80.891,
      "step": 39180
    },
    {
      "epoch": 0.32419241427803286,
      "grad_norm": 1265.29638671875,
      "learning_rate": 8.63662623567024e-06,
      "loss": 100.5664,
      "step": 39190
    },
    {
      "epoch": 0.32427513752740206,
      "grad_norm": 975.9793701171875,
      "learning_rate": 8.63565816605872e-06,
      "loss": 98.3975,
      "step": 39200
    },
    {
      "epoch": 0.3243578607767713,
      "grad_norm": 1260.0736083984375,
      "learning_rate": 8.634689807174564e-06,
      "loss": 122.2016,
      "step": 39210
    },
    {
      "epoch": 0.32444058402614057,
      "grad_norm": 745.4339599609375,
      "learning_rate": 8.633721159094823e-06,
      "loss": 118.936,
      "step": 39220
    },
    {
      "epoch": 0.32452330727550976,
      "grad_norm": 803.7348022460938,
      "learning_rate": 8.632752221896562e-06,
      "loss": 76.6836,
      "step": 39230
    },
    {
      "epoch": 0.324606030524879,
      "grad_norm": 901.3794555664062,
      "learning_rate": 8.631782995656884e-06,
      "loss": 114.3698,
      "step": 39240
    },
    {
      "epoch": 0.3246887537742483,
      "grad_norm": 2110.507080078125,
      "learning_rate": 8.630813480452898e-06,
      "loss": 109.549,
      "step": 39250
    },
    {
      "epoch": 0.32477147702361747,
      "grad_norm": 868.3207397460938,
      "learning_rate": 8.629843676361747e-06,
      "loss": 147.7418,
      "step": 39260
    },
    {
      "epoch": 0.3248542002729867,
      "grad_norm": 776.3412475585938,
      "learning_rate": 8.628873583460593e-06,
      "loss": 126.7695,
      "step": 39270
    },
    {
      "epoch": 0.324936923522356,
      "grad_norm": 1359.6387939453125,
      "learning_rate": 8.627903201826622e-06,
      "loss": 120.8187,
      "step": 39280
    },
    {
      "epoch": 0.3250196467717252,
      "grad_norm": 1282.950439453125,
      "learning_rate": 8.626932531537042e-06,
      "loss": 123.0786,
      "step": 39290
    },
    {
      "epoch": 0.32510237002109443,
      "grad_norm": 530.9898071289062,
      "learning_rate": 8.625961572669087e-06,
      "loss": 120.1885,
      "step": 39300
    },
    {
      "epoch": 0.3251850932704637,
      "grad_norm": 997.0206298828125,
      "learning_rate": 8.62499032530001e-06,
      "loss": 120.9154,
      "step": 39310
    },
    {
      "epoch": 0.3252678165198329,
      "grad_norm": 676.1962890625,
      "learning_rate": 8.624018789507091e-06,
      "loss": 97.3104,
      "step": 39320
    },
    {
      "epoch": 0.32535053976920214,
      "grad_norm": 692.87255859375,
      "learning_rate": 8.62304696536763e-06,
      "loss": 118.6817,
      "step": 39330
    },
    {
      "epoch": 0.3254332630185714,
      "grad_norm": 980.8681030273438,
      "learning_rate": 8.622074852958946e-06,
      "loss": 112.1015,
      "step": 39340
    },
    {
      "epoch": 0.3255159862679406,
      "grad_norm": 1021.6939697265625,
      "learning_rate": 8.621102452358393e-06,
      "loss": 190.7402,
      "step": 39350
    },
    {
      "epoch": 0.32559870951730985,
      "grad_norm": 1223.068115234375,
      "learning_rate": 8.620129763643333e-06,
      "loss": 128.2917,
      "step": 39360
    },
    {
      "epoch": 0.3256814327666791,
      "grad_norm": 1219.437744140625,
      "learning_rate": 8.619156786891162e-06,
      "loss": 136.7339,
      "step": 39370
    },
    {
      "epoch": 0.3257641560160483,
      "grad_norm": 1412.034912109375,
      "learning_rate": 8.618183522179295e-06,
      "loss": 131.6702,
      "step": 39380
    },
    {
      "epoch": 0.32584687926541755,
      "grad_norm": 1203.2330322265625,
      "learning_rate": 8.617209969585171e-06,
      "loss": 88.7958,
      "step": 39390
    },
    {
      "epoch": 0.32592960251478675,
      "grad_norm": 809.6597290039062,
      "learning_rate": 8.616236129186252e-06,
      "loss": 102.6644,
      "step": 39400
    },
    {
      "epoch": 0.326012325764156,
      "grad_norm": 3100.604736328125,
      "learning_rate": 8.615262001060019e-06,
      "loss": 176.8819,
      "step": 39410
    },
    {
      "epoch": 0.32609504901352526,
      "grad_norm": 1474.5286865234375,
      "learning_rate": 8.61428758528398e-06,
      "loss": 123.6517,
      "step": 39420
    },
    {
      "epoch": 0.32617777226289446,
      "grad_norm": 893.0943603515625,
      "learning_rate": 8.613312881935667e-06,
      "loss": 118.6461,
      "step": 39430
    },
    {
      "epoch": 0.3262604955122637,
      "grad_norm": 1108.7327880859375,
      "learning_rate": 8.61233789109263e-06,
      "loss": 129.676,
      "step": 39440
    },
    {
      "epoch": 0.32634321876163297,
      "grad_norm": 799.1575317382812,
      "learning_rate": 8.611362612832445e-06,
      "loss": 109.5865,
      "step": 39450
    },
    {
      "epoch": 0.32642594201100217,
      "grad_norm": 1402.5484619140625,
      "learning_rate": 8.610387047232711e-06,
      "loss": 103.1031,
      "step": 39460
    },
    {
      "epoch": 0.3265086652603714,
      "grad_norm": 1590.6834716796875,
      "learning_rate": 8.609411194371049e-06,
      "loss": 114.6393,
      "step": 39470
    },
    {
      "epoch": 0.3265913885097407,
      "grad_norm": 1389.5260009765625,
      "learning_rate": 8.608435054325103e-06,
      "loss": 100.0405,
      "step": 39480
    },
    {
      "epoch": 0.3266741117591099,
      "grad_norm": 1510.9293212890625,
      "learning_rate": 8.60745862717254e-06,
      "loss": 78.9952,
      "step": 39490
    },
    {
      "epoch": 0.32675683500847913,
      "grad_norm": 2515.405029296875,
      "learning_rate": 8.606481912991052e-06,
      "loss": 125.8343,
      "step": 39500
    },
    {
      "epoch": 0.3268395582578484,
      "grad_norm": 1044.6246337890625,
      "learning_rate": 8.605504911858347e-06,
      "loss": 95.9947,
      "step": 39510
    },
    {
      "epoch": 0.3269222815072176,
      "grad_norm": 742.9393920898438,
      "learning_rate": 8.604527623852165e-06,
      "loss": 129.3403,
      "step": 39520
    },
    {
      "epoch": 0.32700500475658684,
      "grad_norm": 843.3123779296875,
      "learning_rate": 8.603550049050262e-06,
      "loss": 124.6452,
      "step": 39530
    },
    {
      "epoch": 0.3270877280059561,
      "grad_norm": 629.9082641601562,
      "learning_rate": 8.602572187530421e-06,
      "loss": 103.7542,
      "step": 39540
    },
    {
      "epoch": 0.3271704512553253,
      "grad_norm": 1079.95556640625,
      "learning_rate": 8.601594039370441e-06,
      "loss": 117.5058,
      "step": 39550
    },
    {
      "epoch": 0.32725317450469454,
      "grad_norm": 785.5013427734375,
      "learning_rate": 8.600615604648155e-06,
      "loss": 67.8067,
      "step": 39560
    },
    {
      "epoch": 0.3273358977540638,
      "grad_norm": 831.0355224609375,
      "learning_rate": 8.599636883441408e-06,
      "loss": 123.9131,
      "step": 39570
    },
    {
      "epoch": 0.327418621003433,
      "grad_norm": 2647.583251953125,
      "learning_rate": 8.598657875828078e-06,
      "loss": 107.162,
      "step": 39580
    },
    {
      "epoch": 0.32750134425280225,
      "grad_norm": 2026.9219970703125,
      "learning_rate": 8.597678581886055e-06,
      "loss": 130.6936,
      "step": 39590
    },
    {
      "epoch": 0.3275840675021715,
      "grad_norm": 991.1618041992188,
      "learning_rate": 8.596699001693257e-06,
      "loss": 107.1374,
      "step": 39600
    },
    {
      "epoch": 0.3276667907515407,
      "grad_norm": 1521.5604248046875,
      "learning_rate": 8.595719135327627e-06,
      "loss": 98.5977,
      "step": 39610
    },
    {
      "epoch": 0.32774951400090996,
      "grad_norm": 613.502685546875,
      "learning_rate": 8.594738982867126e-06,
      "loss": 100.9653,
      "step": 39620
    },
    {
      "epoch": 0.3278322372502792,
      "grad_norm": 1424.407470703125,
      "learning_rate": 8.593758544389743e-06,
      "loss": 119.1369,
      "step": 39630
    },
    {
      "epoch": 0.3279149604996484,
      "grad_norm": 519.1179809570312,
      "learning_rate": 8.592777819973486e-06,
      "loss": 121.2218,
      "step": 39640
    },
    {
      "epoch": 0.32799768374901767,
      "grad_norm": 1449.661865234375,
      "learning_rate": 8.591796809696386e-06,
      "loss": 114.1455,
      "step": 39650
    },
    {
      "epoch": 0.3280804069983869,
      "grad_norm": 986.8948364257812,
      "learning_rate": 8.590815513636498e-06,
      "loss": 111.6402,
      "step": 39660
    },
    {
      "epoch": 0.3281631302477561,
      "grad_norm": 839.0859375,
      "learning_rate": 8.5898339318719e-06,
      "loss": 85.4794,
      "step": 39670
    },
    {
      "epoch": 0.3282458534971254,
      "grad_norm": 1216.8238525390625,
      "learning_rate": 8.58885206448069e-06,
      "loss": 126.5229,
      "step": 39680
    },
    {
      "epoch": 0.3283285767464946,
      "grad_norm": 1210.4658203125,
      "learning_rate": 8.587869911540993e-06,
      "loss": 131.2425,
      "step": 39690
    },
    {
      "epoch": 0.3284112999958638,
      "grad_norm": 638.7323608398438,
      "learning_rate": 8.586887473130951e-06,
      "loss": 117.0074,
      "step": 39700
    },
    {
      "epoch": 0.3284940232452331,
      "grad_norm": 1674.9326171875,
      "learning_rate": 8.585904749328736e-06,
      "loss": 101.3178,
      "step": 39710
    },
    {
      "epoch": 0.32857674649460233,
      "grad_norm": 730.9718627929688,
      "learning_rate": 8.584921740212537e-06,
      "loss": 79.7682,
      "step": 39720
    },
    {
      "epoch": 0.32865946974397153,
      "grad_norm": 849.2908325195312,
      "learning_rate": 8.583938445860569e-06,
      "loss": 134.0528,
      "step": 39730
    },
    {
      "epoch": 0.3287421929933408,
      "grad_norm": 1976.713134765625,
      "learning_rate": 8.582954866351065e-06,
      "loss": 109.2086,
      "step": 39740
    },
    {
      "epoch": 0.32882491624271004,
      "grad_norm": 1140.69140625,
      "learning_rate": 8.581971001762287e-06,
      "loss": 115.7576,
      "step": 39750
    },
    {
      "epoch": 0.32890763949207924,
      "grad_norm": 1253.4771728515625,
      "learning_rate": 8.580986852172514e-06,
      "loss": 99.3701,
      "step": 39760
    },
    {
      "epoch": 0.3289903627414485,
      "grad_norm": 1577.8370361328125,
      "learning_rate": 8.580002417660054e-06,
      "loss": 137.5488,
      "step": 39770
    },
    {
      "epoch": 0.3290730859908177,
      "grad_norm": 759.9320068359375,
      "learning_rate": 8.579017698303228e-06,
      "loss": 110.6118,
      "step": 39780
    },
    {
      "epoch": 0.32915580924018695,
      "grad_norm": 807.0444946289062,
      "learning_rate": 8.578032694180394e-06,
      "loss": 108.5404,
      "step": 39790
    },
    {
      "epoch": 0.3292385324895562,
      "grad_norm": 901.5609741210938,
      "learning_rate": 8.577047405369916e-06,
      "loss": 92.3528,
      "step": 39800
    },
    {
      "epoch": 0.3293212557389254,
      "grad_norm": 815.9768676757812,
      "learning_rate": 8.576061831950193e-06,
      "loss": 116.8808,
      "step": 39810
    },
    {
      "epoch": 0.32940397898829465,
      "grad_norm": 676.6227416992188,
      "learning_rate": 8.575075973999642e-06,
      "loss": 104.0332,
      "step": 39820
    },
    {
      "epoch": 0.3294867022376639,
      "grad_norm": 655.098876953125,
      "learning_rate": 8.574089831596703e-06,
      "loss": 114.4098,
      "step": 39830
    },
    {
      "epoch": 0.3295694254870331,
      "grad_norm": 1169.68359375,
      "learning_rate": 8.57310340481984e-06,
      "loss": 88.758,
      "step": 39840
    },
    {
      "epoch": 0.32965214873640236,
      "grad_norm": 635.2750244140625,
      "learning_rate": 8.572116693747537e-06,
      "loss": 98.1875,
      "step": 39850
    },
    {
      "epoch": 0.3297348719857716,
      "grad_norm": 797.3588256835938,
      "learning_rate": 8.571129698458302e-06,
      "loss": 101.033,
      "step": 39860
    },
    {
      "epoch": 0.3298175952351408,
      "grad_norm": 1276.1683349609375,
      "learning_rate": 8.570142419030668e-06,
      "loss": 111.7359,
      "step": 39870
    },
    {
      "epoch": 0.32990031848451007,
      "grad_norm": 700.5169677734375,
      "learning_rate": 8.569154855543184e-06,
      "loss": 101.829,
      "step": 39880
    },
    {
      "epoch": 0.3299830417338793,
      "grad_norm": 1665.83984375,
      "learning_rate": 8.56816700807443e-06,
      "loss": 120.2223,
      "step": 39890
    },
    {
      "epoch": 0.3300657649832485,
      "grad_norm": 619.46435546875,
      "learning_rate": 8.567178876703002e-06,
      "loss": 101.4117,
      "step": 39900
    },
    {
      "epoch": 0.3301484882326178,
      "grad_norm": 1413.1790771484375,
      "learning_rate": 8.566190461507521e-06,
      "loss": 108.3938,
      "step": 39910
    },
    {
      "epoch": 0.33023121148198703,
      "grad_norm": 917.2999877929688,
      "learning_rate": 8.565201762566632e-06,
      "loss": 80.6623,
      "step": 39920
    },
    {
      "epoch": 0.33031393473135623,
      "grad_norm": 787.8756713867188,
      "learning_rate": 8.564212779959003e-06,
      "loss": 130.3724,
      "step": 39930
    },
    {
      "epoch": 0.3303966579807255,
      "grad_norm": 649.1900634765625,
      "learning_rate": 8.563223513763319e-06,
      "loss": 107.5673,
      "step": 39940
    },
    {
      "epoch": 0.33047938123009474,
      "grad_norm": 656.0354614257812,
      "learning_rate": 8.562233964058294e-06,
      "loss": 138.9998,
      "step": 39950
    },
    {
      "epoch": 0.33056210447946394,
      "grad_norm": 1630.0479736328125,
      "learning_rate": 8.561244130922658e-06,
      "loss": 79.6873,
      "step": 39960
    },
    {
      "epoch": 0.3306448277288332,
      "grad_norm": 908.9981689453125,
      "learning_rate": 8.560254014435172e-06,
      "loss": 124.0382,
      "step": 39970
    },
    {
      "epoch": 0.33072755097820244,
      "grad_norm": 1071.929931640625,
      "learning_rate": 8.559263614674615e-06,
      "loss": 102.3747,
      "step": 39980
    },
    {
      "epoch": 0.33081027422757164,
      "grad_norm": 827.8428344726562,
      "learning_rate": 8.558272931719785e-06,
      "loss": 100.2324,
      "step": 39990
    },
    {
      "epoch": 0.3308929974769409,
      "grad_norm": 2593.23779296875,
      "learning_rate": 8.557281965649508e-06,
      "loss": 107.4415,
      "step": 40000
    },
    {
      "epoch": 0.33097572072631015,
      "grad_norm": 1010.6577758789062,
      "learning_rate": 8.556290716542632e-06,
      "loss": 84.2611,
      "step": 40010
    },
    {
      "epoch": 0.33105844397567935,
      "grad_norm": 787.0457153320312,
      "learning_rate": 8.555299184478026e-06,
      "loss": 129.8781,
      "step": 40020
    },
    {
      "epoch": 0.3311411672250486,
      "grad_norm": 953.2777099609375,
      "learning_rate": 8.554307369534577e-06,
      "loss": 103.5916,
      "step": 40030
    },
    {
      "epoch": 0.33122389047441786,
      "grad_norm": 1413.6817626953125,
      "learning_rate": 8.553315271791207e-06,
      "loss": 92.5186,
      "step": 40040
    },
    {
      "epoch": 0.33130661372378706,
      "grad_norm": 1188.9342041015625,
      "learning_rate": 8.552322891326846e-06,
      "loss": 98.2379,
      "step": 40050
    },
    {
      "epoch": 0.3313893369731563,
      "grad_norm": 1236.0517578125,
      "learning_rate": 8.551330228220454e-06,
      "loss": 107.0516,
      "step": 40060
    },
    {
      "epoch": 0.33147206022252557,
      "grad_norm": 919.5791015625,
      "learning_rate": 8.550337282551016e-06,
      "loss": 101.3186,
      "step": 40070
    },
    {
      "epoch": 0.33155478347189477,
      "grad_norm": 1503.684326171875,
      "learning_rate": 8.549344054397533e-06,
      "loss": 108.6517,
      "step": 40080
    },
    {
      "epoch": 0.331637506721264,
      "grad_norm": 419.35028076171875,
      "learning_rate": 8.548350543839034e-06,
      "loss": 85.9801,
      "step": 40090
    },
    {
      "epoch": 0.3317202299706333,
      "grad_norm": 1511.6007080078125,
      "learning_rate": 8.547356750954568e-06,
      "loss": 109.9655,
      "step": 40100
    },
    {
      "epoch": 0.3318029532200025,
      "grad_norm": 811.8706665039062,
      "learning_rate": 8.546362675823204e-06,
      "loss": 186.2005,
      "step": 40110
    },
    {
      "epoch": 0.3318856764693717,
      "grad_norm": 910.4181518554688,
      "learning_rate": 8.545368318524036e-06,
      "loss": 113.1569,
      "step": 40120
    },
    {
      "epoch": 0.3319683997187409,
      "grad_norm": 841.266357421875,
      "learning_rate": 8.544373679136184e-06,
      "loss": 102.7385,
      "step": 40130
    },
    {
      "epoch": 0.3320511229681102,
      "grad_norm": 1087.2705078125,
      "learning_rate": 8.543378757738785e-06,
      "loss": 83.0132,
      "step": 40140
    },
    {
      "epoch": 0.33213384621747943,
      "grad_norm": 844.1511840820312,
      "learning_rate": 8.542383554411e-06,
      "loss": 97.4727,
      "step": 40150
    },
    {
      "epoch": 0.33221656946684863,
      "grad_norm": 1138.1463623046875,
      "learning_rate": 8.541388069232012e-06,
      "loss": 96.1207,
      "step": 40160
    },
    {
      "epoch": 0.3322992927162179,
      "grad_norm": 1386.313232421875,
      "learning_rate": 8.54039230228103e-06,
      "loss": 121.9133,
      "step": 40170
    },
    {
      "epoch": 0.33238201596558714,
      "grad_norm": 1644.196044921875,
      "learning_rate": 8.53939625363728e-06,
      "loss": 110.2846,
      "step": 40180
    },
    {
      "epoch": 0.33246473921495634,
      "grad_norm": 663.9628295898438,
      "learning_rate": 8.538399923380011e-06,
      "loss": 147.1378,
      "step": 40190
    },
    {
      "epoch": 0.3325474624643256,
      "grad_norm": 921.0543212890625,
      "learning_rate": 8.537403311588502e-06,
      "loss": 94.7127,
      "step": 40200
    },
    {
      "epoch": 0.33263018571369485,
      "grad_norm": 1512.567626953125,
      "learning_rate": 8.536406418342044e-06,
      "loss": 87.8837,
      "step": 40210
    },
    {
      "epoch": 0.33271290896306405,
      "grad_norm": 1098.78369140625,
      "learning_rate": 8.53540924371996e-06,
      "loss": 108.6445,
      "step": 40220
    },
    {
      "epoch": 0.3327956322124333,
      "grad_norm": 1846.2921142578125,
      "learning_rate": 8.534411787801586e-06,
      "loss": 95.6519,
      "step": 40230
    },
    {
      "epoch": 0.33287835546180256,
      "grad_norm": 1103.0107421875,
      "learning_rate": 8.533414050666287e-06,
      "loss": 109.1561,
      "step": 40240
    },
    {
      "epoch": 0.33296107871117175,
      "grad_norm": 903.0642700195312,
      "learning_rate": 8.532416032393447e-06,
      "loss": 99.2833,
      "step": 40250
    },
    {
      "epoch": 0.333043801960541,
      "grad_norm": 1340.9583740234375,
      "learning_rate": 8.531417733062476e-06,
      "loss": 116.9413,
      "step": 40260
    },
    {
      "epoch": 0.33312652520991026,
      "grad_norm": 1119.4525146484375,
      "learning_rate": 8.530419152752804e-06,
      "loss": 124.4811,
      "step": 40270
    },
    {
      "epoch": 0.33320924845927946,
      "grad_norm": 1566.3560791015625,
      "learning_rate": 8.529420291543882e-06,
      "loss": 127.6215,
      "step": 40280
    },
    {
      "epoch": 0.3332919717086487,
      "grad_norm": 1316.3895263671875,
      "learning_rate": 8.528421149515185e-06,
      "loss": 107.4906,
      "step": 40290
    },
    {
      "epoch": 0.33337469495801797,
      "grad_norm": 825.5098266601562,
      "learning_rate": 8.52742172674621e-06,
      "loss": 129.04,
      "step": 40300
    },
    {
      "epoch": 0.33345741820738717,
      "grad_norm": 1096.2099609375,
      "learning_rate": 8.526422023316478e-06,
      "loss": 91.2496,
      "step": 40310
    },
    {
      "epoch": 0.3335401414567564,
      "grad_norm": 781.985107421875,
      "learning_rate": 8.525422039305529e-06,
      "loss": 124.4936,
      "step": 40320
    },
    {
      "epoch": 0.3336228647061257,
      "grad_norm": 1158.1458740234375,
      "learning_rate": 8.524421774792926e-06,
      "loss": 84.0795,
      "step": 40330
    },
    {
      "epoch": 0.3337055879554949,
      "grad_norm": 1207.7098388671875,
      "learning_rate": 8.52342122985826e-06,
      "loss": 91.0112,
      "step": 40340
    },
    {
      "epoch": 0.33378831120486413,
      "grad_norm": 847.1415405273438,
      "learning_rate": 8.522420404581135e-06,
      "loss": 101.6437,
      "step": 40350
    },
    {
      "epoch": 0.3338710344542334,
      "grad_norm": 995.0086669921875,
      "learning_rate": 8.521419299041185e-06,
      "loss": 93.0817,
      "step": 40360
    },
    {
      "epoch": 0.3339537577036026,
      "grad_norm": 892.8756103515625,
      "learning_rate": 8.520417913318065e-06,
      "loss": 110.2036,
      "step": 40370
    },
    {
      "epoch": 0.33403648095297184,
      "grad_norm": 1183.2650146484375,
      "learning_rate": 8.519416247491445e-06,
      "loss": 126.0844,
      "step": 40380
    },
    {
      "epoch": 0.3341192042023411,
      "grad_norm": 605.9612426757812,
      "learning_rate": 8.518414301641027e-06,
      "loss": 69.7784,
      "step": 40390
    },
    {
      "epoch": 0.3342019274517103,
      "grad_norm": 1190.9295654296875,
      "learning_rate": 8.517412075846529e-06,
      "loss": 138.7514,
      "step": 40400
    },
    {
      "epoch": 0.33428465070107954,
      "grad_norm": 792.2731323242188,
      "learning_rate": 8.516409570187698e-06,
      "loss": 98.212,
      "step": 40410
    },
    {
      "epoch": 0.3343673739504488,
      "grad_norm": 820.3512573242188,
      "learning_rate": 8.515406784744294e-06,
      "loss": 87.1192,
      "step": 40420
    },
    {
      "epoch": 0.334450097199818,
      "grad_norm": 1030.4779052734375,
      "learning_rate": 8.514403719596104e-06,
      "loss": 112.4568,
      "step": 40430
    },
    {
      "epoch": 0.33453282044918725,
      "grad_norm": 649.7733764648438,
      "learning_rate": 8.513400374822942e-06,
      "loss": 129.4392,
      "step": 40440
    },
    {
      "epoch": 0.3346155436985565,
      "grad_norm": 1169.7542724609375,
      "learning_rate": 8.512396750504635e-06,
      "loss": 96.2116,
      "step": 40450
    },
    {
      "epoch": 0.3346982669479257,
      "grad_norm": 1340.5599365234375,
      "learning_rate": 8.511392846721037e-06,
      "loss": 130.6511,
      "step": 40460
    },
    {
      "epoch": 0.33478099019729496,
      "grad_norm": 1084.3248291015625,
      "learning_rate": 8.510388663552027e-06,
      "loss": 96.2522,
      "step": 40470
    },
    {
      "epoch": 0.33486371344666416,
      "grad_norm": 940.4993896484375,
      "learning_rate": 8.509384201077502e-06,
      "loss": 182.6661,
      "step": 40480
    },
    {
      "epoch": 0.3349464366960334,
      "grad_norm": 473.4619445800781,
      "learning_rate": 8.508379459377381e-06,
      "loss": 98.2326,
      "step": 40490
    },
    {
      "epoch": 0.33502915994540267,
      "grad_norm": 915.9678344726562,
      "learning_rate": 8.507374438531606e-06,
      "loss": 90.029,
      "step": 40500
    },
    {
      "epoch": 0.33511188319477186,
      "grad_norm": 644.4666748046875,
      "learning_rate": 8.506369138620148e-06,
      "loss": 160.186,
      "step": 40510
    },
    {
      "epoch": 0.3351946064441411,
      "grad_norm": 771.171875,
      "learning_rate": 8.505363559722985e-06,
      "loss": 96.5032,
      "step": 40520
    },
    {
      "epoch": 0.3352773296935104,
      "grad_norm": 575.1885375976562,
      "learning_rate": 8.504357701920134e-06,
      "loss": 78.7146,
      "step": 40530
    },
    {
      "epoch": 0.33536005294287957,
      "grad_norm": 728.8525390625,
      "learning_rate": 8.503351565291622e-06,
      "loss": 130.1776,
      "step": 40540
    },
    {
      "epoch": 0.3354427761922488,
      "grad_norm": 1049.326416015625,
      "learning_rate": 8.502345149917506e-06,
      "loss": 91.4142,
      "step": 40550
    },
    {
      "epoch": 0.3355254994416181,
      "grad_norm": 1567.1068115234375,
      "learning_rate": 8.501338455877859e-06,
      "loss": 128.1109,
      "step": 40560
    },
    {
      "epoch": 0.3356082226909873,
      "grad_norm": 639.6390380859375,
      "learning_rate": 8.50033148325278e-06,
      "loss": 89.2162,
      "step": 40570
    },
    {
      "epoch": 0.33569094594035653,
      "grad_norm": 1309.4464111328125,
      "learning_rate": 8.499324232122389e-06,
      "loss": 119.5868,
      "step": 40580
    },
    {
      "epoch": 0.3357736691897258,
      "grad_norm": 961.0704345703125,
      "learning_rate": 8.498316702566828e-06,
      "loss": 108.5671,
      "step": 40590
    },
    {
      "epoch": 0.335856392439095,
      "grad_norm": 1294.3653564453125,
      "learning_rate": 8.497308894666263e-06,
      "loss": 114.0025,
      "step": 40600
    },
    {
      "epoch": 0.33593911568846424,
      "grad_norm": 1210.564697265625,
      "learning_rate": 8.496300808500878e-06,
      "loss": 122.3642,
      "step": 40610
    },
    {
      "epoch": 0.3360218389378335,
      "grad_norm": 1512.3568115234375,
      "learning_rate": 8.495292444150887e-06,
      "loss": 146.3031,
      "step": 40620
    },
    {
      "epoch": 0.3361045621872027,
      "grad_norm": 956.16162109375,
      "learning_rate": 8.494283801696514e-06,
      "loss": 140.3855,
      "step": 40630
    },
    {
      "epoch": 0.33618728543657195,
      "grad_norm": 862.5502319335938,
      "learning_rate": 8.493274881218017e-06,
      "loss": 92.3681,
      "step": 40640
    },
    {
      "epoch": 0.3362700086859412,
      "grad_norm": 715.0096435546875,
      "learning_rate": 8.49226568279567e-06,
      "loss": 93.8356,
      "step": 40650
    },
    {
      "epoch": 0.3363527319353104,
      "grad_norm": 858.4218139648438,
      "learning_rate": 8.49125620650977e-06,
      "loss": 89.4689,
      "step": 40660
    },
    {
      "epoch": 0.33643545518467965,
      "grad_norm": 797.36328125,
      "learning_rate": 8.490246452440636e-06,
      "loss": 131.3191,
      "step": 40670
    },
    {
      "epoch": 0.3365181784340489,
      "grad_norm": 887.1189575195312,
      "learning_rate": 8.48923642066861e-06,
      "loss": 143.5478,
      "step": 40680
    },
    {
      "epoch": 0.3366009016834181,
      "grad_norm": 688.75732421875,
      "learning_rate": 8.488226111274055e-06,
      "loss": 129.7013,
      "step": 40690
    },
    {
      "epoch": 0.33668362493278736,
      "grad_norm": 770.3273315429688,
      "learning_rate": 8.487215524337357e-06,
      "loss": 77.0376,
      "step": 40700
    },
    {
      "epoch": 0.3367663481821566,
      "grad_norm": 960.6439819335938,
      "learning_rate": 8.486204659938924e-06,
      "loss": 127.0703,
      "step": 40710
    },
    {
      "epoch": 0.3368490714315258,
      "grad_norm": 764.961669921875,
      "learning_rate": 8.485193518159186e-06,
      "loss": 111.9176,
      "step": 40720
    },
    {
      "epoch": 0.33693179468089507,
      "grad_norm": 900.38134765625,
      "learning_rate": 8.484182099078596e-06,
      "loss": 112.1536,
      "step": 40730
    },
    {
      "epoch": 0.3370145179302643,
      "grad_norm": 960.113525390625,
      "learning_rate": 8.483170402777624e-06,
      "loss": 127.3206,
      "step": 40740
    },
    {
      "epoch": 0.3370972411796335,
      "grad_norm": 1646.762939453125,
      "learning_rate": 8.482158429336769e-06,
      "loss": 143.5467,
      "step": 40750
    },
    {
      "epoch": 0.3371799644290028,
      "grad_norm": 985.1450805664062,
      "learning_rate": 8.48114617883655e-06,
      "loss": 114.6298,
      "step": 40760
    },
    {
      "epoch": 0.33726268767837203,
      "grad_norm": 759.6780395507812,
      "learning_rate": 8.480133651357507e-06,
      "loss": 116.8154,
      "step": 40770
    },
    {
      "epoch": 0.33734541092774123,
      "grad_norm": 885.9656982421875,
      "learning_rate": 8.479120846980197e-06,
      "loss": 108.7685,
      "step": 40780
    },
    {
      "epoch": 0.3374281341771105,
      "grad_norm": 575.7235717773438,
      "learning_rate": 8.478107765785212e-06,
      "loss": 88.1911,
      "step": 40790
    },
    {
      "epoch": 0.33751085742647974,
      "grad_norm": 666.4598388671875,
      "learning_rate": 8.477094407853153e-06,
      "loss": 69.0146,
      "step": 40800
    },
    {
      "epoch": 0.33759358067584894,
      "grad_norm": 645.81201171875,
      "learning_rate": 8.47608077326465e-06,
      "loss": 94.4932,
      "step": 40810
    },
    {
      "epoch": 0.3376763039252182,
      "grad_norm": 994.3779907226562,
      "learning_rate": 8.475066862100352e-06,
      "loss": 88.9415,
      "step": 40820
    },
    {
      "epoch": 0.33775902717458745,
      "grad_norm": 759.6505737304688,
      "learning_rate": 8.474052674440934e-06,
      "loss": 117.1598,
      "step": 40830
    },
    {
      "epoch": 0.33784175042395664,
      "grad_norm": 869.6111450195312,
      "learning_rate": 8.473038210367086e-06,
      "loss": 78.2449,
      "step": 40840
    },
    {
      "epoch": 0.3379244736733259,
      "grad_norm": 947.123046875,
      "learning_rate": 8.47202346995953e-06,
      "loss": 93.2359,
      "step": 40850
    },
    {
      "epoch": 0.3380071969226951,
      "grad_norm": 1361.4940185546875,
      "learning_rate": 8.471008453298998e-06,
      "loss": 134.8301,
      "step": 40860
    },
    {
      "epoch": 0.33808992017206435,
      "grad_norm": 729.060546875,
      "learning_rate": 8.469993160466254e-06,
      "loss": 94.2659,
      "step": 40870
    },
    {
      "epoch": 0.3381726434214336,
      "grad_norm": 643.9381713867188,
      "learning_rate": 8.46897759154208e-06,
      "loss": 88.6729,
      "step": 40880
    },
    {
      "epoch": 0.3382553666708028,
      "grad_norm": 927.8340454101562,
      "learning_rate": 8.467961746607279e-06,
      "loss": 127.7991,
      "step": 40890
    },
    {
      "epoch": 0.33833808992017206,
      "grad_norm": 604.8875122070312,
      "learning_rate": 8.466945625742678e-06,
      "loss": 76.509,
      "step": 40900
    },
    {
      "epoch": 0.3384208131695413,
      "grad_norm": 672.4412841796875,
      "learning_rate": 8.465929229029124e-06,
      "loss": 165.3088,
      "step": 40910
    },
    {
      "epoch": 0.3385035364189105,
      "grad_norm": 649.953369140625,
      "learning_rate": 8.464912556547486e-06,
      "loss": 112.9105,
      "step": 40920
    },
    {
      "epoch": 0.33858625966827977,
      "grad_norm": 1930.8785400390625,
      "learning_rate": 8.46389560837866e-06,
      "loss": 108.5322,
      "step": 40930
    },
    {
      "epoch": 0.338668982917649,
      "grad_norm": 622.740966796875,
      "learning_rate": 8.462878384603558e-06,
      "loss": 117.5824,
      "step": 40940
    },
    {
      "epoch": 0.3387517061670182,
      "grad_norm": 474.0765686035156,
      "learning_rate": 8.461860885303116e-06,
      "loss": 99.5456,
      "step": 40950
    },
    {
      "epoch": 0.3388344294163875,
      "grad_norm": 952.2894287109375,
      "learning_rate": 8.460843110558287e-06,
      "loss": 124.9169,
      "step": 40960
    },
    {
      "epoch": 0.3389171526657567,
      "grad_norm": 654.9668579101562,
      "learning_rate": 8.459825060450058e-06,
      "loss": 90.4174,
      "step": 40970
    },
    {
      "epoch": 0.3389998759151259,
      "grad_norm": 1069.80029296875,
      "learning_rate": 8.458806735059428e-06,
      "loss": 134.0334,
      "step": 40980
    },
    {
      "epoch": 0.3390825991644952,
      "grad_norm": 657.5958862304688,
      "learning_rate": 8.45778813446742e-06,
      "loss": 97.927,
      "step": 40990
    },
    {
      "epoch": 0.33916532241386443,
      "grad_norm": 1148.0079345703125,
      "learning_rate": 8.456769258755078e-06,
      "loss": 111.504,
      "step": 41000
    },
    {
      "epoch": 0.33924804566323363,
      "grad_norm": 1073.7718505859375,
      "learning_rate": 8.455750108003468e-06,
      "loss": 78.7796,
      "step": 41010
    },
    {
      "epoch": 0.3393307689126029,
      "grad_norm": 530.935302734375,
      "learning_rate": 8.454730682293686e-06,
      "loss": 76.4729,
      "step": 41020
    },
    {
      "epoch": 0.33941349216197214,
      "grad_norm": 584.3358764648438,
      "learning_rate": 8.453710981706838e-06,
      "loss": 100.3047,
      "step": 41030
    },
    {
      "epoch": 0.33949621541134134,
      "grad_norm": 1019.848388671875,
      "learning_rate": 8.452691006324055e-06,
      "loss": 101.324,
      "step": 41040
    },
    {
      "epoch": 0.3395789386607106,
      "grad_norm": 991.2681274414062,
      "learning_rate": 8.451670756226496e-06,
      "loss": 75.0817,
      "step": 41050
    },
    {
      "epoch": 0.33966166191007985,
      "grad_norm": 730.0400390625,
      "learning_rate": 8.450650231495336e-06,
      "loss": 85.305,
      "step": 41060
    },
    {
      "epoch": 0.33974438515944905,
      "grad_norm": 929.1596069335938,
      "learning_rate": 8.449629432211774e-06,
      "loss": 92.8536,
      "step": 41070
    },
    {
      "epoch": 0.3398271084088183,
      "grad_norm": 1377.75146484375,
      "learning_rate": 8.44860835845703e-06,
      "loss": 103.3764,
      "step": 41080
    },
    {
      "epoch": 0.33990983165818756,
      "grad_norm": 522.6357421875,
      "learning_rate": 8.447587010312343e-06,
      "loss": 107.57,
      "step": 41090
    },
    {
      "epoch": 0.33999255490755675,
      "grad_norm": 2292.6142578125,
      "learning_rate": 8.44656538785898e-06,
      "loss": 135.6061,
      "step": 41100
    },
    {
      "epoch": 0.340075278156926,
      "grad_norm": 419.8533630371094,
      "learning_rate": 8.44554349117823e-06,
      "loss": 80.6014,
      "step": 41110
    },
    {
      "epoch": 0.34015800140629526,
      "grad_norm": 1163.7928466796875,
      "learning_rate": 8.444521320351397e-06,
      "loss": 110.4075,
      "step": 41120
    },
    {
      "epoch": 0.34024072465566446,
      "grad_norm": 942.8880615234375,
      "learning_rate": 8.44349887545981e-06,
      "loss": 118.3985,
      "step": 41130
    },
    {
      "epoch": 0.3403234479050337,
      "grad_norm": 509.7681884765625,
      "learning_rate": 8.442476156584818e-06,
      "loss": 133.2833,
      "step": 41140
    },
    {
      "epoch": 0.34040617115440297,
      "grad_norm": 1343.72705078125,
      "learning_rate": 8.4414531638078e-06,
      "loss": 134.3201,
      "step": 41150
    },
    {
      "epoch": 0.34048889440377217,
      "grad_norm": 1334.9111328125,
      "learning_rate": 8.440429897210148e-06,
      "loss": 94.3114,
      "step": 41160
    },
    {
      "epoch": 0.3405716176531414,
      "grad_norm": 918.8824462890625,
      "learning_rate": 8.439406356873279e-06,
      "loss": 105.6756,
      "step": 41170
    },
    {
      "epoch": 0.3406543409025107,
      "grad_norm": 435.05010986328125,
      "learning_rate": 8.43838254287863e-06,
      "loss": 86.3829,
      "step": 41180
    },
    {
      "epoch": 0.3407370641518799,
      "grad_norm": 934.824462890625,
      "learning_rate": 8.43735845530766e-06,
      "loss": 122.4491,
      "step": 41190
    },
    {
      "epoch": 0.34081978740124913,
      "grad_norm": 1095.854248046875,
      "learning_rate": 8.436334094241855e-06,
      "loss": 110.3371,
      "step": 41200
    },
    {
      "epoch": 0.34090251065061833,
      "grad_norm": 1048.58154296875,
      "learning_rate": 8.435309459762718e-06,
      "loss": 135.8438,
      "step": 41210
    },
    {
      "epoch": 0.3409852338999876,
      "grad_norm": 722.0227661132812,
      "learning_rate": 8.434284551951772e-06,
      "loss": 86.2307,
      "step": 41220
    },
    {
      "epoch": 0.34106795714935684,
      "grad_norm": 835.3677368164062,
      "learning_rate": 8.433259370890565e-06,
      "loss": 79.5151,
      "step": 41230
    },
    {
      "epoch": 0.34115068039872604,
      "grad_norm": 1341.2940673828125,
      "learning_rate": 8.432233916660669e-06,
      "loss": 102.6455,
      "step": 41240
    },
    {
      "epoch": 0.3412334036480953,
      "grad_norm": 972.4404907226562,
      "learning_rate": 8.43120818934367e-06,
      "loss": 105.9913,
      "step": 41250
    },
    {
      "epoch": 0.34131612689746454,
      "grad_norm": 752.3978881835938,
      "learning_rate": 8.43018218902118e-06,
      "loss": 97.6527,
      "step": 41260
    },
    {
      "epoch": 0.34139885014683374,
      "grad_norm": 744.5557250976562,
      "learning_rate": 8.429155915774839e-06,
      "loss": 98.6538,
      "step": 41270
    },
    {
      "epoch": 0.341481573396203,
      "grad_norm": 1084.3350830078125,
      "learning_rate": 8.428129369686299e-06,
      "loss": 96.4803,
      "step": 41280
    },
    {
      "epoch": 0.34156429664557225,
      "grad_norm": 887.330078125,
      "learning_rate": 8.427102550837238e-06,
      "loss": 111.9868,
      "step": 41290
    },
    {
      "epoch": 0.34164701989494145,
      "grad_norm": 869.30908203125,
      "learning_rate": 8.426075459309356e-06,
      "loss": 120.3469,
      "step": 41300
    },
    {
      "epoch": 0.3417297431443107,
      "grad_norm": 2174.565673828125,
      "learning_rate": 8.42504809518437e-06,
      "loss": 115.0281,
      "step": 41310
    },
    {
      "epoch": 0.34181246639367996,
      "grad_norm": 939.2781372070312,
      "learning_rate": 8.42402045854403e-06,
      "loss": 118.3146,
      "step": 41320
    },
    {
      "epoch": 0.34189518964304916,
      "grad_norm": 962.0004272460938,
      "learning_rate": 8.422992549470094e-06,
      "loss": 111.6336,
      "step": 41330
    },
    {
      "epoch": 0.3419779128924184,
      "grad_norm": 856.4796142578125,
      "learning_rate": 8.42196436804435e-06,
      "loss": 89.8677,
      "step": 41340
    },
    {
      "epoch": 0.34206063614178767,
      "grad_norm": 882.47509765625,
      "learning_rate": 8.420935914348607e-06,
      "loss": 109.2613,
      "step": 41350
    },
    {
      "epoch": 0.34214335939115686,
      "grad_norm": 1152.5184326171875,
      "learning_rate": 8.419907188464691e-06,
      "loss": 83.5429,
      "step": 41360
    },
    {
      "epoch": 0.3422260826405261,
      "grad_norm": 1086.84521484375,
      "learning_rate": 8.418878190474459e-06,
      "loss": 107.8546,
      "step": 41370
    },
    {
      "epoch": 0.3423088058898954,
      "grad_norm": 1217.004150390625,
      "learning_rate": 8.417848920459778e-06,
      "loss": 137.2482,
      "step": 41380
    },
    {
      "epoch": 0.34239152913926457,
      "grad_norm": 706.2305297851562,
      "learning_rate": 8.416819378502543e-06,
      "loss": 74.3434,
      "step": 41390
    },
    {
      "epoch": 0.3424742523886338,
      "grad_norm": 1614.1566162109375,
      "learning_rate": 8.415789564684673e-06,
      "loss": 142.1887,
      "step": 41400
    },
    {
      "epoch": 0.3425569756380031,
      "grad_norm": 875.3898315429688,
      "learning_rate": 8.414759479088102e-06,
      "loss": 97.1488,
      "step": 41410
    },
    {
      "epoch": 0.3426396988873723,
      "grad_norm": 1026.1566162109375,
      "learning_rate": 8.413729121794794e-06,
      "loss": 130.0628,
      "step": 41420
    },
    {
      "epoch": 0.34272242213674153,
      "grad_norm": 977.263427734375,
      "learning_rate": 8.412698492886723e-06,
      "loss": 78.8849,
      "step": 41430
    },
    {
      "epoch": 0.3428051453861108,
      "grad_norm": 1244.2081298828125,
      "learning_rate": 8.411667592445898e-06,
      "loss": 142.179,
      "step": 41440
    },
    {
      "epoch": 0.34288786863548,
      "grad_norm": 751.5814819335938,
      "learning_rate": 8.410636420554337e-06,
      "loss": 95.5533,
      "step": 41450
    },
    {
      "epoch": 0.34297059188484924,
      "grad_norm": 786.3610229492188,
      "learning_rate": 8.409604977294093e-06,
      "loss": 103.3417,
      "step": 41460
    },
    {
      "epoch": 0.3430533151342185,
      "grad_norm": 931.486572265625,
      "learning_rate": 8.408573262747225e-06,
      "loss": 141.6174,
      "step": 41470
    },
    {
      "epoch": 0.3431360383835877,
      "grad_norm": 838.9139404296875,
      "learning_rate": 8.407541276995828e-06,
      "loss": 105.8545,
      "step": 41480
    },
    {
      "epoch": 0.34321876163295695,
      "grad_norm": 654.8321533203125,
      "learning_rate": 8.40650902012201e-06,
      "loss": 107.5087,
      "step": 41490
    },
    {
      "epoch": 0.3433014848823262,
      "grad_norm": 766.4340209960938,
      "learning_rate": 8.405476492207902e-06,
      "loss": 113.7733,
      "step": 41500
    },
    {
      "epoch": 0.3433842081316954,
      "grad_norm": 1260.985595703125,
      "learning_rate": 8.404443693335658e-06,
      "loss": 108.2389,
      "step": 41510
    },
    {
      "epoch": 0.34346693138106466,
      "grad_norm": 777.0344848632812,
      "learning_rate": 8.403410623587454e-06,
      "loss": 112.1793,
      "step": 41520
    },
    {
      "epoch": 0.3435496546304339,
      "grad_norm": 836.8817749023438,
      "learning_rate": 8.402377283045487e-06,
      "loss": 121.241,
      "step": 41530
    },
    {
      "epoch": 0.3436323778798031,
      "grad_norm": 675.0514526367188,
      "learning_rate": 8.401343671791974e-06,
      "loss": 121.9953,
      "step": 41540
    },
    {
      "epoch": 0.34371510112917236,
      "grad_norm": 1098.513916015625,
      "learning_rate": 8.400309789909155e-06,
      "loss": 129.7842,
      "step": 41550
    },
    {
      "epoch": 0.3437978243785416,
      "grad_norm": 728.864013671875,
      "learning_rate": 8.399275637479291e-06,
      "loss": 99.4059,
      "step": 41560
    },
    {
      "epoch": 0.3438805476279108,
      "grad_norm": 744.5280151367188,
      "learning_rate": 8.398241214584666e-06,
      "loss": 115.1609,
      "step": 41570
    },
    {
      "epoch": 0.34396327087728007,
      "grad_norm": 938.07080078125,
      "learning_rate": 8.397206521307584e-06,
      "loss": 102.9427,
      "step": 41580
    },
    {
      "epoch": 0.34404599412664927,
      "grad_norm": 1658.468994140625,
      "learning_rate": 8.396171557730369e-06,
      "loss": 115.574,
      "step": 41590
    },
    {
      "epoch": 0.3441287173760185,
      "grad_norm": 634.4130859375,
      "learning_rate": 8.39513632393537e-06,
      "loss": 93.7718,
      "step": 41600
    },
    {
      "epoch": 0.3442114406253878,
      "grad_norm": 579.6549072265625,
      "learning_rate": 8.394100820004954e-06,
      "loss": 92.9008,
      "step": 41610
    },
    {
      "epoch": 0.344294163874757,
      "grad_norm": 450.8374938964844,
      "learning_rate": 8.393065046021513e-06,
      "loss": 90.1996,
      "step": 41620
    },
    {
      "epoch": 0.34437688712412623,
      "grad_norm": 572.2847290039062,
      "learning_rate": 8.39202900206746e-06,
      "loss": 147.2724,
      "step": 41630
    },
    {
      "epoch": 0.3444596103734955,
      "grad_norm": 2148.787353515625,
      "learning_rate": 8.390992688225226e-06,
      "loss": 132.4656,
      "step": 41640
    },
    {
      "epoch": 0.3445423336228647,
      "grad_norm": 707.6249389648438,
      "learning_rate": 8.389956104577265e-06,
      "loss": 90.7987,
      "step": 41650
    },
    {
      "epoch": 0.34462505687223394,
      "grad_norm": 1235.2393798828125,
      "learning_rate": 8.388919251206054e-06,
      "loss": 120.5997,
      "step": 41660
    },
    {
      "epoch": 0.3447077801216032,
      "grad_norm": 802.1920166015625,
      "learning_rate": 8.387882128194094e-06,
      "loss": 127.8519,
      "step": 41670
    },
    {
      "epoch": 0.3447905033709724,
      "grad_norm": 1075.386962890625,
      "learning_rate": 8.3868447356239e-06,
      "loss": 95.1565,
      "step": 41680
    },
    {
      "epoch": 0.34487322662034164,
      "grad_norm": 1524.40283203125,
      "learning_rate": 8.385807073578014e-06,
      "loss": 125.0543,
      "step": 41690
    },
    {
      "epoch": 0.3449559498697109,
      "grad_norm": 887.3607177734375,
      "learning_rate": 8.384769142138998e-06,
      "loss": 86.6021,
      "step": 41700
    },
    {
      "epoch": 0.3450386731190801,
      "grad_norm": 932.5927124023438,
      "learning_rate": 8.383730941389434e-06,
      "loss": 96.8807,
      "step": 41710
    },
    {
      "epoch": 0.34512139636844935,
      "grad_norm": 956.2900390625,
      "learning_rate": 8.382692471411931e-06,
      "loss": 97.0625,
      "step": 41720
    },
    {
      "epoch": 0.3452041196178186,
      "grad_norm": 646.645263671875,
      "learning_rate": 8.38165373228911e-06,
      "loss": 95.7504,
      "step": 41730
    },
    {
      "epoch": 0.3452868428671878,
      "grad_norm": 1114.9278564453125,
      "learning_rate": 8.380614724103622e-06,
      "loss": 120.9078,
      "step": 41740
    },
    {
      "epoch": 0.34536956611655706,
      "grad_norm": 693.3104248046875,
      "learning_rate": 8.379575446938136e-06,
      "loss": 106.318,
      "step": 41750
    },
    {
      "epoch": 0.3454522893659263,
      "grad_norm": 8694.8857421875,
      "learning_rate": 8.37853590087534e-06,
      "loss": 133.577,
      "step": 41760
    },
    {
      "epoch": 0.3455350126152955,
      "grad_norm": 645.8015747070312,
      "learning_rate": 8.377496085997949e-06,
      "loss": 124.9107,
      "step": 41770
    },
    {
      "epoch": 0.34561773586466477,
      "grad_norm": 2443.57421875,
      "learning_rate": 8.376456002388695e-06,
      "loss": 115.6421,
      "step": 41780
    },
    {
      "epoch": 0.345700459114034,
      "grad_norm": 1185.3404541015625,
      "learning_rate": 8.375415650130332e-06,
      "loss": 103.3099,
      "step": 41790
    },
    {
      "epoch": 0.3457831823634032,
      "grad_norm": 1205.232666015625,
      "learning_rate": 8.37437502930564e-06,
      "loss": 100.6235,
      "step": 41800
    },
    {
      "epoch": 0.3458659056127725,
      "grad_norm": 1242.7322998046875,
      "learning_rate": 8.373334139997409e-06,
      "loss": 131.52,
      "step": 41810
    },
    {
      "epoch": 0.3459486288621417,
      "grad_norm": 747.3458862304688,
      "learning_rate": 8.372292982288463e-06,
      "loss": 118.5125,
      "step": 41820
    },
    {
      "epoch": 0.3460313521115109,
      "grad_norm": 1096.74462890625,
      "learning_rate": 8.371251556261642e-06,
      "loss": 136.9112,
      "step": 41830
    },
    {
      "epoch": 0.3461140753608802,
      "grad_norm": 801.9876708984375,
      "learning_rate": 8.370209861999807e-06,
      "loss": 106.0218,
      "step": 41840
    },
    {
      "epoch": 0.34619679861024943,
      "grad_norm": 909.0194091796875,
      "learning_rate": 8.36916789958584e-06,
      "loss": 103.4507,
      "step": 41850
    },
    {
      "epoch": 0.34627952185961863,
      "grad_norm": 1055.92041015625,
      "learning_rate": 8.368125669102645e-06,
      "loss": 94.5659,
      "step": 41860
    },
    {
      "epoch": 0.3463622451089879,
      "grad_norm": 867.0262451171875,
      "learning_rate": 8.36708317063315e-06,
      "loss": 135.4051,
      "step": 41870
    },
    {
      "epoch": 0.34644496835835714,
      "grad_norm": 871.2109985351562,
      "learning_rate": 8.366040404260298e-06,
      "loss": 99.1085,
      "step": 41880
    },
    {
      "epoch": 0.34652769160772634,
      "grad_norm": 830.19384765625,
      "learning_rate": 8.36499737006706e-06,
      "loss": 76.3817,
      "step": 41890
    },
    {
      "epoch": 0.3466104148570956,
      "grad_norm": 951.84326171875,
      "learning_rate": 8.363954068136424e-06,
      "loss": 119.5681,
      "step": 41900
    },
    {
      "epoch": 0.34669313810646485,
      "grad_norm": 731.0087890625,
      "learning_rate": 8.362910498551402e-06,
      "loss": 114.4868,
      "step": 41910
    },
    {
      "epoch": 0.34677586135583405,
      "grad_norm": 635.1656494140625,
      "learning_rate": 8.361866661395024e-06,
      "loss": 116.8612,
      "step": 41920
    },
    {
      "epoch": 0.3468585846052033,
      "grad_norm": 1068.5445556640625,
      "learning_rate": 8.360822556750345e-06,
      "loss": 91.3164,
      "step": 41930
    },
    {
      "epoch": 0.3469413078545725,
      "grad_norm": 949.19873046875,
      "learning_rate": 8.35977818470044e-06,
      "loss": 104.9618,
      "step": 41940
    },
    {
      "epoch": 0.34702403110394175,
      "grad_norm": 429.62677001953125,
      "learning_rate": 8.358733545328404e-06,
      "loss": 93.7747,
      "step": 41950
    },
    {
      "epoch": 0.347106754353311,
      "grad_norm": 893.44189453125,
      "learning_rate": 8.357688638717354e-06,
      "loss": 106.5521,
      "step": 41960
    },
    {
      "epoch": 0.3471894776026802,
      "grad_norm": 585.1188354492188,
      "learning_rate": 8.356643464950428e-06,
      "loss": 80.4151,
      "step": 41970
    },
    {
      "epoch": 0.34727220085204946,
      "grad_norm": 1215.0146484375,
      "learning_rate": 8.355598024110789e-06,
      "loss": 181.5081,
      "step": 41980
    },
    {
      "epoch": 0.3473549241014187,
      "grad_norm": 1458.9388427734375,
      "learning_rate": 8.354552316281613e-06,
      "loss": 141.7899,
      "step": 41990
    },
    {
      "epoch": 0.3474376473507879,
      "grad_norm": 915.29443359375,
      "learning_rate": 8.353506341546106e-06,
      "loss": 108.375,
      "step": 42000
    },
    {
      "epoch": 0.34752037060015717,
      "grad_norm": 1018.5140991210938,
      "learning_rate": 8.352460099987488e-06,
      "loss": 118.4601,
      "step": 42010
    },
    {
      "epoch": 0.3476030938495264,
      "grad_norm": 1653.6846923828125,
      "learning_rate": 8.351413591689007e-06,
      "loss": 127.7061,
      "step": 42020
    },
    {
      "epoch": 0.3476858170988956,
      "grad_norm": 1303.524169921875,
      "learning_rate": 8.350366816733927e-06,
      "loss": 109.428,
      "step": 42030
    },
    {
      "epoch": 0.3477685403482649,
      "grad_norm": 1039.2083740234375,
      "learning_rate": 8.349319775205536e-06,
      "loss": 120.8401,
      "step": 42040
    },
    {
      "epoch": 0.34785126359763413,
      "grad_norm": 1214.731689453125,
      "learning_rate": 8.34827246718714e-06,
      "loss": 157.5093,
      "step": 42050
    },
    {
      "epoch": 0.34793398684700333,
      "grad_norm": 832.408203125,
      "learning_rate": 8.347224892762072e-06,
      "loss": 106.25,
      "step": 42060
    },
    {
      "epoch": 0.3480167100963726,
      "grad_norm": 1899.1883544921875,
      "learning_rate": 8.346177052013681e-06,
      "loss": 128.2392,
      "step": 42070
    },
    {
      "epoch": 0.34809943334574184,
      "grad_norm": 475.4049072265625,
      "learning_rate": 8.345128945025338e-06,
      "loss": 128.0041,
      "step": 42080
    },
    {
      "epoch": 0.34818215659511104,
      "grad_norm": 483.9809875488281,
      "learning_rate": 8.344080571880438e-06,
      "loss": 92.6426,
      "step": 42090
    },
    {
      "epoch": 0.3482648798444803,
      "grad_norm": 1139.35302734375,
      "learning_rate": 8.343031932662394e-06,
      "loss": 89.6336,
      "step": 42100
    },
    {
      "epoch": 0.34834760309384954,
      "grad_norm": 998.1423950195312,
      "learning_rate": 8.341983027454641e-06,
      "loss": 148.1835,
      "step": 42110
    },
    {
      "epoch": 0.34843032634321874,
      "grad_norm": 1257.747802734375,
      "learning_rate": 8.340933856340637e-06,
      "loss": 116.7931,
      "step": 42120
    },
    {
      "epoch": 0.348513049592588,
      "grad_norm": 1050.6527099609375,
      "learning_rate": 8.339884419403857e-06,
      "loss": 124.4426,
      "step": 42130
    },
    {
      "epoch": 0.34859577284195725,
      "grad_norm": 935.9532470703125,
      "learning_rate": 8.338834716727801e-06,
      "loss": 100.9498,
      "step": 42140
    },
    {
      "epoch": 0.34867849609132645,
      "grad_norm": 1319.49365234375,
      "learning_rate": 8.337784748395992e-06,
      "loss": 89.4219,
      "step": 42150
    },
    {
      "epoch": 0.3487612193406957,
      "grad_norm": 583.6215209960938,
      "learning_rate": 8.336734514491968e-06,
      "loss": 101.7242,
      "step": 42160
    },
    {
      "epoch": 0.34884394259006496,
      "grad_norm": 1093.5703125,
      "learning_rate": 8.335684015099294e-06,
      "loss": 100.3031,
      "step": 42170
    },
    {
      "epoch": 0.34892666583943416,
      "grad_norm": 7167.025390625,
      "learning_rate": 8.33463325030155e-06,
      "loss": 142.965,
      "step": 42180
    },
    {
      "epoch": 0.3490093890888034,
      "grad_norm": 1053.3721923828125,
      "learning_rate": 8.333582220182344e-06,
      "loss": 118.1564,
      "step": 42190
    },
    {
      "epoch": 0.34909211233817267,
      "grad_norm": 1960.47705078125,
      "learning_rate": 8.332530924825297e-06,
      "loss": 119.301,
      "step": 42200
    },
    {
      "epoch": 0.34917483558754187,
      "grad_norm": 973.1529541015625,
      "learning_rate": 8.33147936431406e-06,
      "loss": 131.6417,
      "step": 42210
    },
    {
      "epoch": 0.3492575588369111,
      "grad_norm": 925.6248168945312,
      "learning_rate": 8.3304275387323e-06,
      "loss": 116.9732,
      "step": 42220
    },
    {
      "epoch": 0.3493402820862804,
      "grad_norm": 0.0,
      "learning_rate": 8.329375448163703e-06,
      "loss": 87.3547,
      "step": 42230
    },
    {
      "epoch": 0.3494230053356496,
      "grad_norm": 1005.3858032226562,
      "learning_rate": 8.328323092691985e-06,
      "loss": 104.9806,
      "step": 42240
    },
    {
      "epoch": 0.3495057285850188,
      "grad_norm": 790.18603515625,
      "learning_rate": 8.32727047240087e-06,
      "loss": 141.6189,
      "step": 42250
    },
    {
      "epoch": 0.3495884518343881,
      "grad_norm": 885.5327758789062,
      "learning_rate": 8.326217587374115e-06,
      "loss": 95.2874,
      "step": 42260
    },
    {
      "epoch": 0.3496711750837573,
      "grad_norm": 1333.3345947265625,
      "learning_rate": 8.325164437695493e-06,
      "loss": 110.1591,
      "step": 42270
    },
    {
      "epoch": 0.34975389833312653,
      "grad_norm": 444.2149963378906,
      "learning_rate": 8.324111023448795e-06,
      "loss": 89.4089,
      "step": 42280
    },
    {
      "epoch": 0.3498366215824958,
      "grad_norm": 712.4427490234375,
      "learning_rate": 8.32305734471784e-06,
      "loss": 104.8016,
      "step": 42290
    },
    {
      "epoch": 0.349919344831865,
      "grad_norm": 825.5255737304688,
      "learning_rate": 8.322003401586463e-06,
      "loss": 109.8285,
      "step": 42300
    },
    {
      "epoch": 0.35000206808123424,
      "grad_norm": 674.3945922851562,
      "learning_rate": 8.32094919413852e-06,
      "loss": 118.0758,
      "step": 42310
    },
    {
      "epoch": 0.35008479133060344,
      "grad_norm": 1092.6353759765625,
      "learning_rate": 8.319894722457892e-06,
      "loss": 100.6579,
      "step": 42320
    },
    {
      "epoch": 0.3501675145799727,
      "grad_norm": 791.6581420898438,
      "learning_rate": 8.318839986628477e-06,
      "loss": 96.1517,
      "step": 42330
    },
    {
      "epoch": 0.35025023782934195,
      "grad_norm": 483.9101867675781,
      "learning_rate": 8.317784986734194e-06,
      "loss": 86.3806,
      "step": 42340
    },
    {
      "epoch": 0.35033296107871115,
      "grad_norm": 982.6383056640625,
      "learning_rate": 8.316729722858987e-06,
      "loss": 142.3889,
      "step": 42350
    },
    {
      "epoch": 0.3504156843280804,
      "grad_norm": 1031.0849609375,
      "learning_rate": 8.31567419508682e-06,
      "loss": 89.9149,
      "step": 42360
    },
    {
      "epoch": 0.35049840757744966,
      "grad_norm": 1122.879150390625,
      "learning_rate": 8.31461840350167e-06,
      "loss": 103.2544,
      "step": 42370
    },
    {
      "epoch": 0.35058113082681885,
      "grad_norm": 1027.361328125,
      "learning_rate": 8.313562348187549e-06,
      "loss": 90.0591,
      "step": 42380
    },
    {
      "epoch": 0.3506638540761881,
      "grad_norm": 789.521240234375,
      "learning_rate": 8.312506029228478e-06,
      "loss": 90.2937,
      "step": 42390
    },
    {
      "epoch": 0.35074657732555736,
      "grad_norm": 616.3732299804688,
      "learning_rate": 8.311449446708506e-06,
      "loss": 129.6206,
      "step": 42400
    },
    {
      "epoch": 0.35082930057492656,
      "grad_norm": 1325.065185546875,
      "learning_rate": 8.310392600711698e-06,
      "loss": 114.9959,
      "step": 42410
    },
    {
      "epoch": 0.3509120238242958,
      "grad_norm": 976.2049560546875,
      "learning_rate": 8.309335491322143e-06,
      "loss": 95.8001,
      "step": 42420
    },
    {
      "epoch": 0.35099474707366507,
      "grad_norm": 1060.470947265625,
      "learning_rate": 8.30827811862395e-06,
      "loss": 101.8484,
      "step": 42430
    },
    {
      "epoch": 0.35107747032303427,
      "grad_norm": 813.407470703125,
      "learning_rate": 8.307220482701251e-06,
      "loss": 119.0382,
      "step": 42440
    },
    {
      "epoch": 0.3511601935724035,
      "grad_norm": 769.6190795898438,
      "learning_rate": 8.306162583638197e-06,
      "loss": 116.1655,
      "step": 42450
    },
    {
      "epoch": 0.3512429168217728,
      "grad_norm": 1182.1982421875,
      "learning_rate": 8.305104421518959e-06,
      "loss": 116.6159,
      "step": 42460
    },
    {
      "epoch": 0.351325640071142,
      "grad_norm": 1151.38671875,
      "learning_rate": 8.30404599642773e-06,
      "loss": 97.3802,
      "step": 42470
    },
    {
      "epoch": 0.35140836332051123,
      "grad_norm": 0.0,
      "learning_rate": 8.302987308448724e-06,
      "loss": 81.0704,
      "step": 42480
    },
    {
      "epoch": 0.3514910865698805,
      "grad_norm": 728.1912231445312,
      "learning_rate": 8.301928357666178e-06,
      "loss": 92.2258,
      "step": 42490
    },
    {
      "epoch": 0.3515738098192497,
      "grad_norm": 793.99267578125,
      "learning_rate": 8.300869144164346e-06,
      "loss": 110.2738,
      "step": 42500
    },
    {
      "epoch": 0.35165653306861894,
      "grad_norm": 993.7640991210938,
      "learning_rate": 8.299809668027505e-06,
      "loss": 131.7156,
      "step": 42510
    },
    {
      "epoch": 0.3517392563179882,
      "grad_norm": 606.397705078125,
      "learning_rate": 8.298749929339953e-06,
      "loss": 104.0031,
      "step": 42520
    },
    {
      "epoch": 0.3518219795673574,
      "grad_norm": 827.4598388671875,
      "learning_rate": 8.297689928186009e-06,
      "loss": 110.1917,
      "step": 42530
    },
    {
      "epoch": 0.35190470281672664,
      "grad_norm": 2409.9150390625,
      "learning_rate": 8.29662966465001e-06,
      "loss": 135.4566,
      "step": 42540
    },
    {
      "epoch": 0.3519874260660959,
      "grad_norm": 613.3368530273438,
      "learning_rate": 8.295569138816319e-06,
      "loss": 80.187,
      "step": 42550
    },
    {
      "epoch": 0.3520701493154651,
      "grad_norm": 843.9635009765625,
      "learning_rate": 8.294508350769315e-06,
      "loss": 109.0294,
      "step": 42560
    },
    {
      "epoch": 0.35215287256483435,
      "grad_norm": 622.92578125,
      "learning_rate": 8.293447300593402e-06,
      "loss": 127.0855,
      "step": 42570
    },
    {
      "epoch": 0.3522355958142036,
      "grad_norm": 657.6416625976562,
      "learning_rate": 8.292385988373005e-06,
      "loss": 108.607,
      "step": 42580
    },
    {
      "epoch": 0.3523183190635728,
      "grad_norm": 1057.9915771484375,
      "learning_rate": 8.29132441419256e-06,
      "loss": 117.3586,
      "step": 42590
    },
    {
      "epoch": 0.35240104231294206,
      "grad_norm": 704.32421875,
      "learning_rate": 8.290262578136541e-06,
      "loss": 92.2817,
      "step": 42600
    },
    {
      "epoch": 0.3524837655623113,
      "grad_norm": 391.53265380859375,
      "learning_rate": 8.289200480289426e-06,
      "loss": 91.4089,
      "step": 42610
    },
    {
      "epoch": 0.3525664888116805,
      "grad_norm": 1248.0340576171875,
      "learning_rate": 8.288138120735726e-06,
      "loss": 94.6713,
      "step": 42620
    },
    {
      "epoch": 0.35264921206104977,
      "grad_norm": 1576.25,
      "learning_rate": 8.287075499559965e-06,
      "loss": 120.6687,
      "step": 42630
    },
    {
      "epoch": 0.352731935310419,
      "grad_norm": 850.7510375976562,
      "learning_rate": 8.286012616846693e-06,
      "loss": 68.9104,
      "step": 42640
    },
    {
      "epoch": 0.3528146585597882,
      "grad_norm": 1018.4036254882812,
      "learning_rate": 8.284949472680477e-06,
      "loss": 96.3005,
      "step": 42650
    },
    {
      "epoch": 0.3528973818091575,
      "grad_norm": 769.276611328125,
      "learning_rate": 8.283886067145908e-06,
      "loss": 101.5941,
      "step": 42660
    },
    {
      "epoch": 0.35298010505852667,
      "grad_norm": 1009.304931640625,
      "learning_rate": 8.282822400327595e-06,
      "loss": 100.8695,
      "step": 42670
    },
    {
      "epoch": 0.3530628283078959,
      "grad_norm": 965.6119384765625,
      "learning_rate": 8.28175847231017e-06,
      "loss": 112.0248,
      "step": 42680
    },
    {
      "epoch": 0.3531455515572652,
      "grad_norm": 862.6695556640625,
      "learning_rate": 8.280694283178285e-06,
      "loss": 97.2944,
      "step": 42690
    },
    {
      "epoch": 0.3532282748066344,
      "grad_norm": 698.0172119140625,
      "learning_rate": 8.27962983301661e-06,
      "loss": 66.0908,
      "step": 42700
    },
    {
      "epoch": 0.35331099805600363,
      "grad_norm": 680.3596801757812,
      "learning_rate": 8.278565121909845e-06,
      "loss": 56.0012,
      "step": 42710
    },
    {
      "epoch": 0.3533937213053729,
      "grad_norm": 979.0956420898438,
      "learning_rate": 8.277500149942697e-06,
      "loss": 111.6114,
      "step": 42720
    },
    {
      "epoch": 0.3534764445547421,
      "grad_norm": 839.169921875,
      "learning_rate": 8.276434917199904e-06,
      "loss": 126.5481,
      "step": 42730
    },
    {
      "epoch": 0.35355916780411134,
      "grad_norm": 1183.0323486328125,
      "learning_rate": 8.275369423766222e-06,
      "loss": 97.7488,
      "step": 42740
    },
    {
      "epoch": 0.3536418910534806,
      "grad_norm": 1349.1572265625,
      "learning_rate": 8.274303669726427e-06,
      "loss": 110.0713,
      "step": 42750
    },
    {
      "epoch": 0.3537246143028498,
      "grad_norm": 916.60107421875,
      "learning_rate": 8.273237655165314e-06,
      "loss": 106.8208,
      "step": 42760
    },
    {
      "epoch": 0.35380733755221905,
      "grad_norm": 800.146484375,
      "learning_rate": 8.272171380167705e-06,
      "loss": 83.875,
      "step": 42770
    },
    {
      "epoch": 0.3538900608015883,
      "grad_norm": 1539.4642333984375,
      "learning_rate": 8.271104844818436e-06,
      "loss": 130.0894,
      "step": 42780
    },
    {
      "epoch": 0.3539727840509575,
      "grad_norm": 1235.24560546875,
      "learning_rate": 8.270038049202366e-06,
      "loss": 131.9467,
      "step": 42790
    },
    {
      "epoch": 0.35405550730032675,
      "grad_norm": 962.0623168945312,
      "learning_rate": 8.268970993404377e-06,
      "loss": 100.8265,
      "step": 42800
    },
    {
      "epoch": 0.354138230549696,
      "grad_norm": 836.7759399414062,
      "learning_rate": 8.267903677509368e-06,
      "loss": 119.7814,
      "step": 42810
    },
    {
      "epoch": 0.3542209537990652,
      "grad_norm": 1205.632080078125,
      "learning_rate": 8.266836101602263e-06,
      "loss": 107.5889,
      "step": 42820
    },
    {
      "epoch": 0.35430367704843446,
      "grad_norm": 1180.9609375,
      "learning_rate": 8.265768265767999e-06,
      "loss": 146.9007,
      "step": 42830
    },
    {
      "epoch": 0.3543864002978037,
      "grad_norm": 784.8786010742188,
      "learning_rate": 8.264700170091543e-06,
      "loss": 93.1217,
      "step": 42840
    },
    {
      "epoch": 0.3544691235471729,
      "grad_norm": 583.6641845703125,
      "learning_rate": 8.263631814657879e-06,
      "loss": 127.952,
      "step": 42850
    },
    {
      "epoch": 0.35455184679654217,
      "grad_norm": 1015.6278076171875,
      "learning_rate": 8.262563199552007e-06,
      "loss": 109.5995,
      "step": 42860
    },
    {
      "epoch": 0.3546345700459114,
      "grad_norm": 1008.2879638671875,
      "learning_rate": 8.261494324858956e-06,
      "loss": 83.4442,
      "step": 42870
    },
    {
      "epoch": 0.3547172932952806,
      "grad_norm": 934.4283447265625,
      "learning_rate": 8.26042519066377e-06,
      "loss": 81.4526,
      "step": 42880
    },
    {
      "epoch": 0.3548000165446499,
      "grad_norm": 602.0402221679688,
      "learning_rate": 8.259355797051515e-06,
      "loss": 87.8508,
      "step": 42890
    },
    {
      "epoch": 0.35488273979401913,
      "grad_norm": 952.8324584960938,
      "learning_rate": 8.258286144107277e-06,
      "loss": 101.1536,
      "step": 42900
    },
    {
      "epoch": 0.35496546304338833,
      "grad_norm": 924.0335693359375,
      "learning_rate": 8.257216231916162e-06,
      "loss": 113.0049,
      "step": 42910
    },
    {
      "epoch": 0.3550481862927576,
      "grad_norm": 684.9584350585938,
      "learning_rate": 8.256146060563304e-06,
      "loss": 117.6641,
      "step": 42920
    },
    {
      "epoch": 0.35513090954212684,
      "grad_norm": 1416.010009765625,
      "learning_rate": 8.255075630133847e-06,
      "loss": 93.2686,
      "step": 42930
    },
    {
      "epoch": 0.35521363279149604,
      "grad_norm": 1231.5831298828125,
      "learning_rate": 8.254004940712958e-06,
      "loss": 111.2918,
      "step": 42940
    },
    {
      "epoch": 0.3552963560408653,
      "grad_norm": 1368.33935546875,
      "learning_rate": 8.252933992385833e-06,
      "loss": 101.9154,
      "step": 42950
    },
    {
      "epoch": 0.35537907929023455,
      "grad_norm": 754.49609375,
      "learning_rate": 8.251862785237676e-06,
      "loss": 88.8121,
      "step": 42960
    },
    {
      "epoch": 0.35546180253960374,
      "grad_norm": 1223.3607177734375,
      "learning_rate": 8.250791319353723e-06,
      "loss": 111.0358,
      "step": 42970
    },
    {
      "epoch": 0.355544525788973,
      "grad_norm": 1084.2374267578125,
      "learning_rate": 8.249719594819225e-06,
      "loss": 107.8028,
      "step": 42980
    },
    {
      "epoch": 0.35562724903834225,
      "grad_norm": 1002.8360595703125,
      "learning_rate": 8.248647611719452e-06,
      "loss": 87.2639,
      "step": 42990
    },
    {
      "epoch": 0.35570997228771145,
      "grad_norm": 882.518310546875,
      "learning_rate": 8.247575370139695e-06,
      "loss": 120.7826,
      "step": 43000
    },
    {
      "epoch": 0.3557926955370807,
      "grad_norm": 710.4674682617188,
      "learning_rate": 8.246502870165273e-06,
      "loss": 130.1348,
      "step": 43010
    },
    {
      "epoch": 0.3558754187864499,
      "grad_norm": 705.7244262695312,
      "learning_rate": 8.245430111881519e-06,
      "loss": 82.2953,
      "step": 43020
    },
    {
      "epoch": 0.35595814203581916,
      "grad_norm": 1012.916748046875,
      "learning_rate": 8.244357095373783e-06,
      "loss": 103.8642,
      "step": 43030
    },
    {
      "epoch": 0.3560408652851884,
      "grad_norm": 823.5924682617188,
      "learning_rate": 8.243283820727441e-06,
      "loss": 120.169,
      "step": 43040
    },
    {
      "epoch": 0.3561235885345576,
      "grad_norm": 506.7729187011719,
      "learning_rate": 8.242210288027893e-06,
      "loss": 106.6605,
      "step": 43050
    },
    {
      "epoch": 0.35620631178392687,
      "grad_norm": 573.303955078125,
      "learning_rate": 8.241136497360552e-06,
      "loss": 96.0248,
      "step": 43060
    },
    {
      "epoch": 0.3562890350332961,
      "grad_norm": 1349.73974609375,
      "learning_rate": 8.240062448810853e-06,
      "loss": 150.305,
      "step": 43070
    },
    {
      "epoch": 0.3563717582826653,
      "grad_norm": 898.83349609375,
      "learning_rate": 8.238988142464254e-06,
      "loss": 99.9782,
      "step": 43080
    },
    {
      "epoch": 0.3564544815320346,
      "grad_norm": 1892.2440185546875,
      "learning_rate": 8.237913578406236e-06,
      "loss": 127.4689,
      "step": 43090
    },
    {
      "epoch": 0.3565372047814038,
      "grad_norm": 1101.9765625,
      "learning_rate": 8.236838756722294e-06,
      "loss": 99.9242,
      "step": 43100
    },
    {
      "epoch": 0.356619928030773,
      "grad_norm": 785.5177001953125,
      "learning_rate": 8.235763677497945e-06,
      "loss": 102.3918,
      "step": 43110
    },
    {
      "epoch": 0.3567026512801423,
      "grad_norm": 1374.7742919921875,
      "learning_rate": 8.234688340818732e-06,
      "loss": 113.4264,
      "step": 43120
    },
    {
      "epoch": 0.35678537452951153,
      "grad_norm": 606.4083862304688,
      "learning_rate": 8.233612746770214e-06,
      "loss": 93.4949,
      "step": 43130
    },
    {
      "epoch": 0.35686809777888073,
      "grad_norm": 922.9923095703125,
      "learning_rate": 8.232536895437968e-06,
      "loss": 85.8221,
      "step": 43140
    },
    {
      "epoch": 0.35695082102825,
      "grad_norm": 790.5139770507812,
      "learning_rate": 8.231460786907597e-06,
      "loss": 117.3538,
      "step": 43150
    },
    {
      "epoch": 0.35703354427761924,
      "grad_norm": 417.8067626953125,
      "learning_rate": 8.230384421264722e-06,
      "loss": 87.6937,
      "step": 43160
    },
    {
      "epoch": 0.35711626752698844,
      "grad_norm": 1458.1962890625,
      "learning_rate": 8.229307798594985e-06,
      "loss": 133.923,
      "step": 43170
    },
    {
      "epoch": 0.3571989907763577,
      "grad_norm": 987.9276123046875,
      "learning_rate": 8.228230918984046e-06,
      "loss": 111.6961,
      "step": 43180
    },
    {
      "epoch": 0.35728171402572695,
      "grad_norm": 1103.388427734375,
      "learning_rate": 8.22715378251759e-06,
      "loss": 93.5427,
      "step": 43190
    },
    {
      "epoch": 0.35736443727509615,
      "grad_norm": 697.0567626953125,
      "learning_rate": 8.226076389281316e-06,
      "loss": 117.8876,
      "step": 43200
    },
    {
      "epoch": 0.3574471605244654,
      "grad_norm": 1062.5294189453125,
      "learning_rate": 8.22499873936095e-06,
      "loss": 85.2779,
      "step": 43210
    },
    {
      "epoch": 0.35752988377383466,
      "grad_norm": 997.5350341796875,
      "learning_rate": 8.223920832842236e-06,
      "loss": 127.3359,
      "step": 43220
    },
    {
      "epoch": 0.35761260702320385,
      "grad_norm": 839.9952392578125,
      "learning_rate": 8.222842669810936e-06,
      "loss": 112.6368,
      "step": 43230
    },
    {
      "epoch": 0.3576953302725731,
      "grad_norm": 1012.621337890625,
      "learning_rate": 8.221764250352835e-06,
      "loss": 108.84,
      "step": 43240
    },
    {
      "epoch": 0.35777805352194236,
      "grad_norm": 918.2972412109375,
      "learning_rate": 8.220685574553739e-06,
      "loss": 85.8282,
      "step": 43250
    },
    {
      "epoch": 0.35786077677131156,
      "grad_norm": 776.9376220703125,
      "learning_rate": 8.219606642499474e-06,
      "loss": 96.5936,
      "step": 43260
    },
    {
      "epoch": 0.3579435000206808,
      "grad_norm": 868.3795166015625,
      "learning_rate": 8.218527454275884e-06,
      "loss": 91.7565,
      "step": 43270
    },
    {
      "epoch": 0.35802622327005007,
      "grad_norm": 958.1405029296875,
      "learning_rate": 8.217448009968834e-06,
      "loss": 110.3028,
      "step": 43280
    },
    {
      "epoch": 0.35810894651941927,
      "grad_norm": 1635.0120849609375,
      "learning_rate": 8.216368309664213e-06,
      "loss": 115.6983,
      "step": 43290
    },
    {
      "epoch": 0.3581916697687885,
      "grad_norm": 495.8813781738281,
      "learning_rate": 8.215288353447927e-06,
      "loss": 125.6738,
      "step": 43300
    },
    {
      "epoch": 0.3582743930181578,
      "grad_norm": 1934.987548828125,
      "learning_rate": 8.214208141405903e-06,
      "loss": 96.0109,
      "step": 43310
    },
    {
      "epoch": 0.358357116267527,
      "grad_norm": 1259.3150634765625,
      "learning_rate": 8.213127673624088e-06,
      "loss": 98.7009,
      "step": 43320
    },
    {
      "epoch": 0.35843983951689623,
      "grad_norm": 985.3812255859375,
      "learning_rate": 8.212046950188451e-06,
      "loss": 119.01,
      "step": 43330
    },
    {
      "epoch": 0.3585225627662655,
      "grad_norm": 657.3631591796875,
      "learning_rate": 8.21096597118498e-06,
      "loss": 102.065,
      "step": 43340
    },
    {
      "epoch": 0.3586052860156347,
      "grad_norm": 808.63232421875,
      "learning_rate": 8.209884736699681e-06,
      "loss": 86.2247,
      "step": 43350
    },
    {
      "epoch": 0.35868800926500394,
      "grad_norm": 963.5828857421875,
      "learning_rate": 8.208803246818586e-06,
      "loss": 99.5541,
      "step": 43360
    },
    {
      "epoch": 0.3587707325143732,
      "grad_norm": 356.1247253417969,
      "learning_rate": 8.207721501627743e-06,
      "loss": 132.9291,
      "step": 43370
    },
    {
      "epoch": 0.3588534557637424,
      "grad_norm": 1359.7357177734375,
      "learning_rate": 8.20663950121322e-06,
      "loss": 106.6871,
      "step": 43380
    },
    {
      "epoch": 0.35893617901311164,
      "grad_norm": 741.9186401367188,
      "learning_rate": 8.20555724566111e-06,
      "loss": 110.5265,
      "step": 43390
    },
    {
      "epoch": 0.35901890226248084,
      "grad_norm": 804.3803100585938,
      "learning_rate": 8.204474735057522e-06,
      "loss": 89.7678,
      "step": 43400
    },
    {
      "epoch": 0.3591016255118501,
      "grad_norm": 986.8497924804688,
      "learning_rate": 8.203391969488586e-06,
      "loss": 76.5805,
      "step": 43410
    },
    {
      "epoch": 0.35918434876121935,
      "grad_norm": 981.0845947265625,
      "learning_rate": 8.20230894904045e-06,
      "loss": 108.0519,
      "step": 43420
    },
    {
      "epoch": 0.35926707201058855,
      "grad_norm": 967.68310546875,
      "learning_rate": 8.20122567379929e-06,
      "loss": 144.2405,
      "step": 43430
    },
    {
      "epoch": 0.3593497952599578,
      "grad_norm": 1253.8336181640625,
      "learning_rate": 8.200142143851295e-06,
      "loss": 85.0357,
      "step": 43440
    },
    {
      "epoch": 0.35943251850932706,
      "grad_norm": 660.7843627929688,
      "learning_rate": 8.199058359282675e-06,
      "loss": 110.5242,
      "step": 43450
    },
    {
      "epoch": 0.35951524175869626,
      "grad_norm": 876.0247802734375,
      "learning_rate": 8.197974320179664e-06,
      "loss": 143.5727,
      "step": 43460
    },
    {
      "epoch": 0.3595979650080655,
      "grad_norm": 1184.567626953125,
      "learning_rate": 8.19689002662851e-06,
      "loss": 87.2803,
      "step": 43470
    },
    {
      "epoch": 0.35968068825743477,
      "grad_norm": 899.4736938476562,
      "learning_rate": 8.195805478715492e-06,
      "loss": 94.5841,
      "step": 43480
    },
    {
      "epoch": 0.35976341150680397,
      "grad_norm": 1176.9891357421875,
      "learning_rate": 8.194720676526898e-06,
      "loss": 105.5688,
      "step": 43490
    },
    {
      "epoch": 0.3598461347561732,
      "grad_norm": 1430.493896484375,
      "learning_rate": 8.193635620149041e-06,
      "loss": 120.161,
      "step": 43500
    },
    {
      "epoch": 0.3599288580055425,
      "grad_norm": 1398.2462158203125,
      "learning_rate": 8.192550309668254e-06,
      "loss": 153.4543,
      "step": 43510
    },
    {
      "epoch": 0.3600115812549117,
      "grad_norm": 978.09033203125,
      "learning_rate": 8.191464745170892e-06,
      "loss": 97.1732,
      "step": 43520
    },
    {
      "epoch": 0.3600943045042809,
      "grad_norm": 1013.3282470703125,
      "learning_rate": 8.190378926743327e-06,
      "loss": 101.2923,
      "step": 43530
    },
    {
      "epoch": 0.3601770277536502,
      "grad_norm": 881.2088012695312,
      "learning_rate": 8.189292854471953e-06,
      "loss": 148.854,
      "step": 43540
    },
    {
      "epoch": 0.3602597510030194,
      "grad_norm": 870.5662841796875,
      "learning_rate": 8.188206528443182e-06,
      "loss": 92.8082,
      "step": 43550
    },
    {
      "epoch": 0.36034247425238863,
      "grad_norm": 776.38916015625,
      "learning_rate": 8.18711994874345e-06,
      "loss": 82.2745,
      "step": 43560
    },
    {
      "epoch": 0.3604251975017579,
      "grad_norm": 904.0106811523438,
      "learning_rate": 8.186033115459211e-06,
      "loss": 97.3916,
      "step": 43570
    },
    {
      "epoch": 0.3605079207511271,
      "grad_norm": 1215.83349609375,
      "learning_rate": 8.184946028676937e-06,
      "loss": 106.3127,
      "step": 43580
    },
    {
      "epoch": 0.36059064400049634,
      "grad_norm": 794.0036010742188,
      "learning_rate": 8.183858688483126e-06,
      "loss": 91.0681,
      "step": 43590
    },
    {
      "epoch": 0.3606733672498656,
      "grad_norm": 1268.9676513671875,
      "learning_rate": 8.182771094964292e-06,
      "loss": 123.4264,
      "step": 43600
    },
    {
      "epoch": 0.3607560904992348,
      "grad_norm": 777.9105224609375,
      "learning_rate": 8.181683248206968e-06,
      "loss": 111.6841,
      "step": 43610
    },
    {
      "epoch": 0.36083881374860405,
      "grad_norm": 854.5106201171875,
      "learning_rate": 8.180595148297709e-06,
      "loss": 113.4441,
      "step": 43620
    },
    {
      "epoch": 0.3609215369979733,
      "grad_norm": 850.8244018554688,
      "learning_rate": 8.179506795323092e-06,
      "loss": 135.2171,
      "step": 43630
    },
    {
      "epoch": 0.3610042602473425,
      "grad_norm": 1269.437255859375,
      "learning_rate": 8.17841818936971e-06,
      "loss": 155.0749,
      "step": 43640
    },
    {
      "epoch": 0.36108698349671176,
      "grad_norm": 1103.447509765625,
      "learning_rate": 8.177329330524182e-06,
      "loss": 85.1156,
      "step": 43650
    },
    {
      "epoch": 0.361169706746081,
      "grad_norm": 800.2313232421875,
      "learning_rate": 8.17624021887314e-06,
      "loss": 103.1669,
      "step": 43660
    },
    {
      "epoch": 0.3612524299954502,
      "grad_norm": 662.4274291992188,
      "learning_rate": 8.17515085450324e-06,
      "loss": 98.6518,
      "step": 43670
    },
    {
      "epoch": 0.36133515324481946,
      "grad_norm": 975.4820556640625,
      "learning_rate": 8.174061237501159e-06,
      "loss": 120.7466,
      "step": 43680
    },
    {
      "epoch": 0.3614178764941887,
      "grad_norm": 1076.44873046875,
      "learning_rate": 8.172971367953593e-06,
      "loss": 80.2128,
      "step": 43690
    },
    {
      "epoch": 0.3615005997435579,
      "grad_norm": 1007.4608764648438,
      "learning_rate": 8.171881245947257e-06,
      "loss": 62.0215,
      "step": 43700
    },
    {
      "epoch": 0.36158332299292717,
      "grad_norm": 1051.32177734375,
      "learning_rate": 8.170790871568887e-06,
      "loss": 157.7504,
      "step": 43710
    },
    {
      "epoch": 0.3616660462422964,
      "grad_norm": 1688.2108154296875,
      "learning_rate": 8.169700244905239e-06,
      "loss": 123.6984,
      "step": 43720
    },
    {
      "epoch": 0.3617487694916656,
      "grad_norm": 1116.3714599609375,
      "learning_rate": 8.168609366043089e-06,
      "loss": 92.2827,
      "step": 43730
    },
    {
      "epoch": 0.3618314927410349,
      "grad_norm": 845.093505859375,
      "learning_rate": 8.167518235069234e-06,
      "loss": 77.9922,
      "step": 43740
    },
    {
      "epoch": 0.3619142159904041,
      "grad_norm": 2042.611572265625,
      "learning_rate": 8.16642685207049e-06,
      "loss": 141.047,
      "step": 43750
    },
    {
      "epoch": 0.36199693923977333,
      "grad_norm": 602.3466186523438,
      "learning_rate": 8.165335217133695e-06,
      "loss": 122.7751,
      "step": 43760
    },
    {
      "epoch": 0.3620796624891426,
      "grad_norm": 998.31005859375,
      "learning_rate": 8.164243330345702e-06,
      "loss": 95.6849,
      "step": 43770
    },
    {
      "epoch": 0.3621623857385118,
      "grad_norm": 939.9814453125,
      "learning_rate": 8.16315119179339e-06,
      "loss": 88.2418,
      "step": 43780
    },
    {
      "epoch": 0.36224510898788104,
      "grad_norm": 2236.7392578125,
      "learning_rate": 8.162058801563652e-06,
      "loss": 116.7937,
      "step": 43790
    },
    {
      "epoch": 0.3623278322372503,
      "grad_norm": 973.8187255859375,
      "learning_rate": 8.160966159743411e-06,
      "loss": 94.5988,
      "step": 43800
    },
    {
      "epoch": 0.3624105554866195,
      "grad_norm": 640.3901977539062,
      "learning_rate": 8.159873266419598e-06,
      "loss": 103.685,
      "step": 43810
    },
    {
      "epoch": 0.36249327873598874,
      "grad_norm": 1432.554931640625,
      "learning_rate": 8.15878012167917e-06,
      "loss": 106.0658,
      "step": 43820
    },
    {
      "epoch": 0.362576001985358,
      "grad_norm": 859.1890258789062,
      "learning_rate": 8.157686725609105e-06,
      "loss": 87.8233,
      "step": 43830
    },
    {
      "epoch": 0.3626587252347272,
      "grad_norm": 1776.3751220703125,
      "learning_rate": 8.1565930782964e-06,
      "loss": 114.2181,
      "step": 43840
    },
    {
      "epoch": 0.36274144848409645,
      "grad_norm": 926.181884765625,
      "learning_rate": 8.155499179828068e-06,
      "loss": 114.7968,
      "step": 43850
    },
    {
      "epoch": 0.3628241717334657,
      "grad_norm": 959.3102416992188,
      "learning_rate": 8.15440503029115e-06,
      "loss": 105.2323,
      "step": 43860
    },
    {
      "epoch": 0.3629068949828349,
      "grad_norm": 1929.3040771484375,
      "learning_rate": 8.153310629772702e-06,
      "loss": 131.4064,
      "step": 43870
    },
    {
      "epoch": 0.36298961823220416,
      "grad_norm": 1120.273193359375,
      "learning_rate": 8.152215978359796e-06,
      "loss": 92.3281,
      "step": 43880
    },
    {
      "epoch": 0.3630723414815734,
      "grad_norm": 830.0736694335938,
      "learning_rate": 8.151121076139534e-06,
      "loss": 91.5073,
      "step": 43890
    },
    {
      "epoch": 0.3631550647309426,
      "grad_norm": 1354.7823486328125,
      "learning_rate": 8.150025923199027e-06,
      "loss": 201.7689,
      "step": 43900
    },
    {
      "epoch": 0.36323778798031187,
      "grad_norm": 1229.7574462890625,
      "learning_rate": 8.148930519625417e-06,
      "loss": 116.0604,
      "step": 43910
    },
    {
      "epoch": 0.3633205112296811,
      "grad_norm": 1157.707763671875,
      "learning_rate": 8.147834865505855e-06,
      "loss": 118.8252,
      "step": 43920
    },
    {
      "epoch": 0.3634032344790503,
      "grad_norm": 391.3431396484375,
      "learning_rate": 8.14673896092752e-06,
      "loss": 94.0042,
      "step": 43930
    },
    {
      "epoch": 0.3634859577284196,
      "grad_norm": 963.4109497070312,
      "learning_rate": 8.145642805977608e-06,
      "loss": 94.306,
      "step": 43940
    },
    {
      "epoch": 0.3635686809777888,
      "grad_norm": 680.21826171875,
      "learning_rate": 8.144546400743334e-06,
      "loss": 121.9921,
      "step": 43950
    },
    {
      "epoch": 0.363651404227158,
      "grad_norm": 763.9702758789062,
      "learning_rate": 8.143449745311934e-06,
      "loss": 105.5048,
      "step": 43960
    },
    {
      "epoch": 0.3637341274765273,
      "grad_norm": 706.5518188476562,
      "learning_rate": 8.142352839770663e-06,
      "loss": 112.0056,
      "step": 43970
    },
    {
      "epoch": 0.36381685072589653,
      "grad_norm": 1533.0765380859375,
      "learning_rate": 8.1412556842068e-06,
      "loss": 109.1279,
      "step": 43980
    },
    {
      "epoch": 0.36389957397526573,
      "grad_norm": 784.7587890625,
      "learning_rate": 8.140158278707637e-06,
      "loss": 121.243,
      "step": 43990
    },
    {
      "epoch": 0.363982297224635,
      "grad_norm": 886.728759765625,
      "learning_rate": 8.139060623360494e-06,
      "loss": 131.5882,
      "step": 44000
    },
    {
      "epoch": 0.36406502047400424,
      "grad_norm": 990.3309936523438,
      "learning_rate": 8.1379627182527e-06,
      "loss": 95.6152,
      "step": 44010
    },
    {
      "epoch": 0.36414774372337344,
      "grad_norm": 1236.8133544921875,
      "learning_rate": 8.136864563471617e-06,
      "loss": 104.638,
      "step": 44020
    },
    {
      "epoch": 0.3642304669727427,
      "grad_norm": 560.0664672851562,
      "learning_rate": 8.135766159104615e-06,
      "loss": 90.362,
      "step": 44030
    },
    {
      "epoch": 0.36431319022211195,
      "grad_norm": 796.653076171875,
      "learning_rate": 8.134667505239092e-06,
      "loss": 99.4829,
      "step": 44040
    },
    {
      "epoch": 0.36439591347148115,
      "grad_norm": 777.6304321289062,
      "learning_rate": 8.133568601962462e-06,
      "loss": 106.3731,
      "step": 44050
    },
    {
      "epoch": 0.3644786367208504,
      "grad_norm": 746.3777465820312,
      "learning_rate": 8.132469449362158e-06,
      "loss": 101.6638,
      "step": 44060
    },
    {
      "epoch": 0.36456135997021966,
      "grad_norm": 824.9530029296875,
      "learning_rate": 8.131370047525637e-06,
      "loss": 144.5076,
      "step": 44070
    },
    {
      "epoch": 0.36464408321958885,
      "grad_norm": 1749.48095703125,
      "learning_rate": 8.130270396540372e-06,
      "loss": 107.8925,
      "step": 44080
    },
    {
      "epoch": 0.3647268064689581,
      "grad_norm": 1590.762451171875,
      "learning_rate": 8.129170496493857e-06,
      "loss": 129.4328,
      "step": 44090
    },
    {
      "epoch": 0.36480952971832736,
      "grad_norm": 854.1180419921875,
      "learning_rate": 8.128070347473609e-06,
      "loss": 97.3397,
      "step": 44100
    },
    {
      "epoch": 0.36489225296769656,
      "grad_norm": 920.0262451171875,
      "learning_rate": 8.126969949567157e-06,
      "loss": 90.8626,
      "step": 44110
    },
    {
      "epoch": 0.3649749762170658,
      "grad_norm": 802.9644775390625,
      "learning_rate": 8.125869302862058e-06,
      "loss": 101.2598,
      "step": 44120
    },
    {
      "epoch": 0.365057699466435,
      "grad_norm": 721.7289428710938,
      "learning_rate": 8.124768407445883e-06,
      "loss": 91.293,
      "step": 44130
    },
    {
      "epoch": 0.36514042271580427,
      "grad_norm": 452.2812805175781,
      "learning_rate": 8.123667263406228e-06,
      "loss": 115.1237,
      "step": 44140
    },
    {
      "epoch": 0.3652231459651735,
      "grad_norm": 755.5316162109375,
      "learning_rate": 8.122565870830704e-06,
      "loss": 95.5803,
      "step": 44150
    },
    {
      "epoch": 0.3653058692145427,
      "grad_norm": 583.1863403320312,
      "learning_rate": 8.121464229806944e-06,
      "loss": 91.2347,
      "step": 44160
    },
    {
      "epoch": 0.365388592463912,
      "grad_norm": 894.2572631835938,
      "learning_rate": 8.120362340422601e-06,
      "loss": 101.122,
      "step": 44170
    },
    {
      "epoch": 0.36547131571328123,
      "grad_norm": 1000.3782958984375,
      "learning_rate": 8.119260202765347e-06,
      "loss": 104.094,
      "step": 44180
    },
    {
      "epoch": 0.36555403896265043,
      "grad_norm": 1205.3262939453125,
      "learning_rate": 8.118157816922874e-06,
      "loss": 107.1193,
      "step": 44190
    },
    {
      "epoch": 0.3656367622120197,
      "grad_norm": 1496.134033203125,
      "learning_rate": 8.117055182982895e-06,
      "loss": 84.7695,
      "step": 44200
    },
    {
      "epoch": 0.36571948546138894,
      "grad_norm": 1437.298095703125,
      "learning_rate": 8.115952301033141e-06,
      "loss": 117.0865,
      "step": 44210
    },
    {
      "epoch": 0.36580220871075814,
      "grad_norm": 1193.9093017578125,
      "learning_rate": 8.11484917116136e-06,
      "loss": 104.2912,
      "step": 44220
    },
    {
      "epoch": 0.3658849319601274,
      "grad_norm": 912.3463745117188,
      "learning_rate": 8.113745793455328e-06,
      "loss": 105.9879,
      "step": 44230
    },
    {
      "epoch": 0.36596765520949665,
      "grad_norm": 1175.63134765625,
      "learning_rate": 8.112642168002831e-06,
      "loss": 95.1146,
      "step": 44240
    },
    {
      "epoch": 0.36605037845886584,
      "grad_norm": 957.2208862304688,
      "learning_rate": 8.111538294891684e-06,
      "loss": 133.3301,
      "step": 44250
    },
    {
      "epoch": 0.3661331017082351,
      "grad_norm": 780.9307861328125,
      "learning_rate": 8.110434174209714e-06,
      "loss": 112.0869,
      "step": 44260
    },
    {
      "epoch": 0.36621582495760435,
      "grad_norm": 1046.54052734375,
      "learning_rate": 8.109329806044772e-06,
      "loss": 122.4268,
      "step": 44270
    },
    {
      "epoch": 0.36629854820697355,
      "grad_norm": 593.833984375,
      "learning_rate": 8.108225190484728e-06,
      "loss": 134.9322,
      "step": 44280
    },
    {
      "epoch": 0.3663812714563428,
      "grad_norm": 1084.296630859375,
      "learning_rate": 8.107120327617469e-06,
      "loss": 108.1544,
      "step": 44290
    },
    {
      "epoch": 0.36646399470571206,
      "grad_norm": 795.716552734375,
      "learning_rate": 8.106015217530906e-06,
      "loss": 104.3661,
      "step": 44300
    },
    {
      "epoch": 0.36654671795508126,
      "grad_norm": 969.0634765625,
      "learning_rate": 8.104909860312968e-06,
      "loss": 118.4515,
      "step": 44310
    },
    {
      "epoch": 0.3666294412044505,
      "grad_norm": 1363.8905029296875,
      "learning_rate": 8.1038042560516e-06,
      "loss": 109.1337,
      "step": 44320
    },
    {
      "epoch": 0.36671216445381977,
      "grad_norm": 715.2791748046875,
      "learning_rate": 8.102698404834773e-06,
      "loss": 79.2838,
      "step": 44330
    },
    {
      "epoch": 0.36679488770318897,
      "grad_norm": 1469.4854736328125,
      "learning_rate": 8.101592306750472e-06,
      "loss": 110.5569,
      "step": 44340
    },
    {
      "epoch": 0.3668776109525582,
      "grad_norm": 992.7304077148438,
      "learning_rate": 8.100485961886707e-06,
      "loss": 97.823,
      "step": 44350
    },
    {
      "epoch": 0.3669603342019275,
      "grad_norm": 1063.874755859375,
      "learning_rate": 8.099379370331502e-06,
      "loss": 112.1215,
      "step": 44360
    },
    {
      "epoch": 0.3670430574512967,
      "grad_norm": 726.4131469726562,
      "learning_rate": 8.098272532172906e-06,
      "loss": 135.1273,
      "step": 44370
    },
    {
      "epoch": 0.3671257807006659,
      "grad_norm": 1017.98193359375,
      "learning_rate": 8.097165447498985e-06,
      "loss": 102.8711,
      "step": 44380
    },
    {
      "epoch": 0.3672085039500352,
      "grad_norm": 1006.7951049804688,
      "learning_rate": 8.09605811639782e-06,
      "loss": 120.6296,
      "step": 44390
    },
    {
      "epoch": 0.3672912271994044,
      "grad_norm": 1553.7777099609375,
      "learning_rate": 8.094950538957523e-06,
      "loss": 116.9153,
      "step": 44400
    },
    {
      "epoch": 0.36737395044877363,
      "grad_norm": 578.3179931640625,
      "learning_rate": 8.093842715266214e-06,
      "loss": 86.2257,
      "step": 44410
    },
    {
      "epoch": 0.3674566736981429,
      "grad_norm": 1568.3966064453125,
      "learning_rate": 8.092734645412037e-06,
      "loss": 103.4828,
      "step": 44420
    },
    {
      "epoch": 0.3675393969475121,
      "grad_norm": 1186.5718994140625,
      "learning_rate": 8.09162632948316e-06,
      "loss": 117.7764,
      "step": 44430
    },
    {
      "epoch": 0.36762212019688134,
      "grad_norm": 1135.4630126953125,
      "learning_rate": 8.090517767567765e-06,
      "loss": 95.9603,
      "step": 44440
    },
    {
      "epoch": 0.3677048434462506,
      "grad_norm": 959.4806518554688,
      "learning_rate": 8.089408959754055e-06,
      "loss": 99.0822,
      "step": 44450
    },
    {
      "epoch": 0.3677875666956198,
      "grad_norm": 758.1735229492188,
      "learning_rate": 8.088299906130252e-06,
      "loss": 149.9401,
      "step": 44460
    },
    {
      "epoch": 0.36787028994498905,
      "grad_norm": 791.1016845703125,
      "learning_rate": 8.087190606784598e-06,
      "loss": 79.0925,
      "step": 44470
    },
    {
      "epoch": 0.36795301319435825,
      "grad_norm": 640.9752807617188,
      "learning_rate": 8.086081061805357e-06,
      "loss": 95.8233,
      "step": 44480
    },
    {
      "epoch": 0.3680357364437275,
      "grad_norm": 1189.6053466796875,
      "learning_rate": 8.084971271280808e-06,
      "loss": 122.035,
      "step": 44490
    },
    {
      "epoch": 0.36811845969309676,
      "grad_norm": 1168.3837890625,
      "learning_rate": 8.083861235299253e-06,
      "loss": 100.2019,
      "step": 44500
    },
    {
      "epoch": 0.36820118294246595,
      "grad_norm": 891.6553955078125,
      "learning_rate": 8.082750953949015e-06,
      "loss": 115.5503,
      "step": 44510
    },
    {
      "epoch": 0.3682839061918352,
      "grad_norm": 884.3746337890625,
      "learning_rate": 8.081640427318429e-06,
      "loss": 94.7234,
      "step": 44520
    },
    {
      "epoch": 0.36836662944120446,
      "grad_norm": 760.7252807617188,
      "learning_rate": 8.080529655495856e-06,
      "loss": 126.0468,
      "step": 44530
    },
    {
      "epoch": 0.36844935269057366,
      "grad_norm": 831.2738647460938,
      "learning_rate": 8.079418638569679e-06,
      "loss": 92.954,
      "step": 44540
    },
    {
      "epoch": 0.3685320759399429,
      "grad_norm": 460.5909118652344,
      "learning_rate": 8.078307376628292e-06,
      "loss": 83.0053,
      "step": 44550
    },
    {
      "epoch": 0.36861479918931217,
      "grad_norm": 795.8884887695312,
      "learning_rate": 8.077195869760114e-06,
      "loss": 86.7317,
      "step": 44560
    },
    {
      "epoch": 0.36869752243868137,
      "grad_norm": 683.8246459960938,
      "learning_rate": 8.076084118053584e-06,
      "loss": 90.0042,
      "step": 44570
    },
    {
      "epoch": 0.3687802456880506,
      "grad_norm": 1078.8155517578125,
      "learning_rate": 8.074972121597158e-06,
      "loss": 140.497,
      "step": 44580
    },
    {
      "epoch": 0.3688629689374199,
      "grad_norm": 447.01214599609375,
      "learning_rate": 8.073859880479314e-06,
      "loss": 76.7604,
      "step": 44590
    },
    {
      "epoch": 0.3689456921867891,
      "grad_norm": 948.76416015625,
      "learning_rate": 8.072747394788545e-06,
      "loss": 83.8023,
      "step": 44600
    },
    {
      "epoch": 0.36902841543615833,
      "grad_norm": 1048.3646240234375,
      "learning_rate": 8.071634664613367e-06,
      "loss": 102.7217,
      "step": 44610
    },
    {
      "epoch": 0.3691111386855276,
      "grad_norm": 859.9012451171875,
      "learning_rate": 8.070521690042317e-06,
      "loss": 103.1552,
      "step": 44620
    },
    {
      "epoch": 0.3691938619348968,
      "grad_norm": 1066.72314453125,
      "learning_rate": 8.069408471163947e-06,
      "loss": 121.4919,
      "step": 44630
    },
    {
      "epoch": 0.36927658518426604,
      "grad_norm": 969.5633544921875,
      "learning_rate": 8.068295008066832e-06,
      "loss": 128.1989,
      "step": 44640
    },
    {
      "epoch": 0.3693593084336353,
      "grad_norm": 1198.6636962890625,
      "learning_rate": 8.067181300839565e-06,
      "loss": 161.2369,
      "step": 44650
    },
    {
      "epoch": 0.3694420316830045,
      "grad_norm": 700.932861328125,
      "learning_rate": 8.066067349570757e-06,
      "loss": 86.4731,
      "step": 44660
    },
    {
      "epoch": 0.36952475493237374,
      "grad_norm": 1984.68017578125,
      "learning_rate": 8.064953154349042e-06,
      "loss": 85.2434,
      "step": 44670
    },
    {
      "epoch": 0.369607478181743,
      "grad_norm": 951.8640747070312,
      "learning_rate": 8.063838715263072e-06,
      "loss": 92.2627,
      "step": 44680
    },
    {
      "epoch": 0.3696902014311122,
      "grad_norm": 991.0081176757812,
      "learning_rate": 8.062724032401515e-06,
      "loss": 82.2411,
      "step": 44690
    },
    {
      "epoch": 0.36977292468048145,
      "grad_norm": 915.6588134765625,
      "learning_rate": 8.061609105853062e-06,
      "loss": 123.3313,
      "step": 44700
    },
    {
      "epoch": 0.3698556479298507,
      "grad_norm": 648.7603759765625,
      "learning_rate": 8.060493935706425e-06,
      "loss": 86.3172,
      "step": 44710
    },
    {
      "epoch": 0.3699383711792199,
      "grad_norm": 731.9179077148438,
      "learning_rate": 8.059378522050332e-06,
      "loss": 142.6297,
      "step": 44720
    },
    {
      "epoch": 0.37002109442858916,
      "grad_norm": 738.2081909179688,
      "learning_rate": 8.05826286497353e-06,
      "loss": 100.2493,
      "step": 44730
    },
    {
      "epoch": 0.3701038176779584,
      "grad_norm": 961.7689208984375,
      "learning_rate": 8.057146964564786e-06,
      "loss": 108.2104,
      "step": 44740
    },
    {
      "epoch": 0.3701865409273276,
      "grad_norm": 1143.8248291015625,
      "learning_rate": 8.05603082091289e-06,
      "loss": 102.1352,
      "step": 44750
    },
    {
      "epoch": 0.37026926417669687,
      "grad_norm": 968.6405639648438,
      "learning_rate": 8.054914434106647e-06,
      "loss": 89.4334,
      "step": 44760
    },
    {
      "epoch": 0.3703519874260661,
      "grad_norm": 800.0516967773438,
      "learning_rate": 8.053797804234882e-06,
      "loss": 100.4505,
      "step": 44770
    },
    {
      "epoch": 0.3704347106754353,
      "grad_norm": 928.8748168945312,
      "learning_rate": 8.052680931386441e-06,
      "loss": 88.3113,
      "step": 44780
    },
    {
      "epoch": 0.3705174339248046,
      "grad_norm": 680.3980102539062,
      "learning_rate": 8.051563815650187e-06,
      "loss": 93.7541,
      "step": 44790
    },
    {
      "epoch": 0.3706001571741738,
      "grad_norm": 988.8343505859375,
      "learning_rate": 8.050446457115005e-06,
      "loss": 138.652,
      "step": 44800
    },
    {
      "epoch": 0.370682880423543,
      "grad_norm": 5196.66015625,
      "learning_rate": 8.0493288558698e-06,
      "loss": 161.6813,
      "step": 44810
    },
    {
      "epoch": 0.3707656036729123,
      "grad_norm": 1195.2532958984375,
      "learning_rate": 8.04821101200349e-06,
      "loss": 104.2872,
      "step": 44820
    },
    {
      "epoch": 0.37084832692228153,
      "grad_norm": 916.4719848632812,
      "learning_rate": 8.047092925605022e-06,
      "loss": 72.5952,
      "step": 44830
    },
    {
      "epoch": 0.37093105017165073,
      "grad_norm": 1935.22119140625,
      "learning_rate": 8.045974596763352e-06,
      "loss": 124.3693,
      "step": 44840
    },
    {
      "epoch": 0.37101377342102,
      "grad_norm": 660.0169677734375,
      "learning_rate": 8.044856025567464e-06,
      "loss": 118.5706,
      "step": 44850
    },
    {
      "epoch": 0.3710964966703892,
      "grad_norm": 1358.344482421875,
      "learning_rate": 8.043737212106356e-06,
      "loss": 116.8551,
      "step": 44860
    },
    {
      "epoch": 0.37117921991975844,
      "grad_norm": 963.0214233398438,
      "learning_rate": 8.042618156469045e-06,
      "loss": 88.5938,
      "step": 44870
    },
    {
      "epoch": 0.3712619431691277,
      "grad_norm": 718.8231811523438,
      "learning_rate": 8.041498858744572e-06,
      "loss": 93.5438,
      "step": 44880
    },
    {
      "epoch": 0.3713446664184969,
      "grad_norm": 884.6295776367188,
      "learning_rate": 8.040379319021994e-06,
      "loss": 98.0878,
      "step": 44890
    },
    {
      "epoch": 0.37142738966786615,
      "grad_norm": 818.326171875,
      "learning_rate": 8.039259537390388e-06,
      "loss": 106.2791,
      "step": 44900
    },
    {
      "epoch": 0.3715101129172354,
      "grad_norm": 722.222900390625,
      "learning_rate": 8.038139513938847e-06,
      "loss": 101.4262,
      "step": 44910
    },
    {
      "epoch": 0.3715928361666046,
      "grad_norm": 706.7052001953125,
      "learning_rate": 8.037019248756488e-06,
      "loss": 103.0186,
      "step": 44920
    },
    {
      "epoch": 0.37167555941597386,
      "grad_norm": 783.2863159179688,
      "learning_rate": 8.035898741932447e-06,
      "loss": 101.9469,
      "step": 44930
    },
    {
      "epoch": 0.3717582826653431,
      "grad_norm": 1484.9461669921875,
      "learning_rate": 8.034777993555875e-06,
      "loss": 140.325,
      "step": 44940
    },
    {
      "epoch": 0.3718410059147123,
      "grad_norm": 669.56640625,
      "learning_rate": 8.033657003715945e-06,
      "loss": 116.9738,
      "step": 44950
    },
    {
      "epoch": 0.37192372916408156,
      "grad_norm": 705.1654663085938,
      "learning_rate": 8.032535772501851e-06,
      "loss": 102.4115,
      "step": 44960
    },
    {
      "epoch": 0.3720064524134508,
      "grad_norm": 1263.1771240234375,
      "learning_rate": 8.031414300002802e-06,
      "loss": 77.7587,
      "step": 44970
    },
    {
      "epoch": 0.37208917566282,
      "grad_norm": 1212.223876953125,
      "learning_rate": 8.03029258630803e-06,
      "loss": 112.9885,
      "step": 44980
    },
    {
      "epoch": 0.37217189891218927,
      "grad_norm": 951.40185546875,
      "learning_rate": 8.029170631506785e-06,
      "loss": 95.1384,
      "step": 44990
    },
    {
      "epoch": 0.3722546221615585,
      "grad_norm": 1134.679443359375,
      "learning_rate": 8.028048435688333e-06,
      "loss": 106.5822,
      "step": 45000
    },
    {
      "epoch": 0.3723373454109277,
      "grad_norm": 932.267822265625,
      "learning_rate": 8.026925998941965e-06,
      "loss": 110.1278,
      "step": 45010
    },
    {
      "epoch": 0.372420068660297,
      "grad_norm": 457.553955078125,
      "learning_rate": 8.025803321356989e-06,
      "loss": 95.6591,
      "step": 45020
    },
    {
      "epoch": 0.37250279190966623,
      "grad_norm": 655.8544921875,
      "learning_rate": 8.024680403022726e-06,
      "loss": 89.4721,
      "step": 45030
    },
    {
      "epoch": 0.37258551515903543,
      "grad_norm": 893.6047973632812,
      "learning_rate": 8.023557244028526e-06,
      "loss": 101.3761,
      "step": 45040
    },
    {
      "epoch": 0.3726682384084047,
      "grad_norm": 765.0892333984375,
      "learning_rate": 8.022433844463752e-06,
      "loss": 85.2344,
      "step": 45050
    },
    {
      "epoch": 0.37275096165777394,
      "grad_norm": 955.836181640625,
      "learning_rate": 8.02131020441779e-06,
      "loss": 105.1853,
      "step": 45060
    },
    {
      "epoch": 0.37283368490714314,
      "grad_norm": 1159.584228515625,
      "learning_rate": 8.02018632398004e-06,
      "loss": 82.6549,
      "step": 45070
    },
    {
      "epoch": 0.3729164081565124,
      "grad_norm": 827.6931762695312,
      "learning_rate": 8.019062203239923e-06,
      "loss": 145.6601,
      "step": 45080
    },
    {
      "epoch": 0.37299913140588165,
      "grad_norm": 1674.2239990234375,
      "learning_rate": 8.017937842286882e-06,
      "loss": 104.1544,
      "step": 45090
    },
    {
      "epoch": 0.37308185465525084,
      "grad_norm": 1010.4392700195312,
      "learning_rate": 8.01681324121038e-06,
      "loss": 102.7155,
      "step": 45100
    },
    {
      "epoch": 0.3731645779046201,
      "grad_norm": 812.0460205078125,
      "learning_rate": 8.015688400099893e-06,
      "loss": 80.9767,
      "step": 45110
    },
    {
      "epoch": 0.37324730115398935,
      "grad_norm": 676.9262084960938,
      "learning_rate": 8.014563319044919e-06,
      "loss": 107.2131,
      "step": 45120
    },
    {
      "epoch": 0.37333002440335855,
      "grad_norm": 1148.935302734375,
      "learning_rate": 8.013437998134978e-06,
      "loss": 96.8717,
      "step": 45130
    },
    {
      "epoch": 0.3734127476527278,
      "grad_norm": 2210.831298828125,
      "learning_rate": 8.012312437459604e-06,
      "loss": 115.9179,
      "step": 45140
    },
    {
      "epoch": 0.37349547090209706,
      "grad_norm": 740.3560180664062,
      "learning_rate": 8.011186637108354e-06,
      "loss": 95.7675,
      "step": 45150
    },
    {
      "epoch": 0.37357819415146626,
      "grad_norm": 1315.1302490234375,
      "learning_rate": 8.010060597170805e-06,
      "loss": 108.1328,
      "step": 45160
    },
    {
      "epoch": 0.3736609174008355,
      "grad_norm": 1144.85400390625,
      "learning_rate": 8.008934317736546e-06,
      "loss": 104.8319,
      "step": 45170
    },
    {
      "epoch": 0.37374364065020477,
      "grad_norm": 642.925048828125,
      "learning_rate": 8.007807798895195e-06,
      "loss": 114.467,
      "step": 45180
    },
    {
      "epoch": 0.37382636389957397,
      "grad_norm": 1433.939208984375,
      "learning_rate": 8.00668104073638e-06,
      "loss": 142.9769,
      "step": 45190
    },
    {
      "epoch": 0.3739090871489432,
      "grad_norm": 615.9909057617188,
      "learning_rate": 8.005554043349753e-06,
      "loss": 77.9182,
      "step": 45200
    },
    {
      "epoch": 0.3739918103983124,
      "grad_norm": 1355.69287109375,
      "learning_rate": 8.004426806824985e-06,
      "loss": 98.0649,
      "step": 45210
    },
    {
      "epoch": 0.3740745336476817,
      "grad_norm": 716.5189208984375,
      "learning_rate": 8.003299331251764e-06,
      "loss": 113.2009,
      "step": 45220
    },
    {
      "epoch": 0.3741572568970509,
      "grad_norm": 860.948974609375,
      "learning_rate": 8.002171616719798e-06,
      "loss": 89.8221,
      "step": 45230
    },
    {
      "epoch": 0.3742399801464201,
      "grad_norm": 1081.593994140625,
      "learning_rate": 8.001043663318815e-06,
      "loss": 148.4487,
      "step": 45240
    },
    {
      "epoch": 0.3743227033957894,
      "grad_norm": 714.3857421875,
      "learning_rate": 7.999915471138562e-06,
      "loss": 111.338,
      "step": 45250
    },
    {
      "epoch": 0.37440542664515863,
      "grad_norm": 863.04052734375,
      "learning_rate": 7.9987870402688e-06,
      "loss": 123.2091,
      "step": 45260
    },
    {
      "epoch": 0.37448814989452783,
      "grad_norm": 541.889404296875,
      "learning_rate": 7.997658370799318e-06,
      "loss": 122.7542,
      "step": 45270
    },
    {
      "epoch": 0.3745708731438971,
      "grad_norm": 948.584716796875,
      "learning_rate": 7.996529462819915e-06,
      "loss": 119.426,
      "step": 45280
    },
    {
      "epoch": 0.37465359639326634,
      "grad_norm": 736.9681396484375,
      "learning_rate": 7.995400316420416e-06,
      "loss": 64.0782,
      "step": 45290
    },
    {
      "epoch": 0.37473631964263554,
      "grad_norm": 1551.1883544921875,
      "learning_rate": 7.994270931690662e-06,
      "loss": 136.7847,
      "step": 45300
    },
    {
      "epoch": 0.3748190428920048,
      "grad_norm": 804.4282836914062,
      "learning_rate": 7.993141308720511e-06,
      "loss": 136.6521,
      "step": 45310
    },
    {
      "epoch": 0.37490176614137405,
      "grad_norm": 778.236083984375,
      "learning_rate": 7.99201144759984e-06,
      "loss": 95.8208,
      "step": 45320
    },
    {
      "epoch": 0.37498448939074325,
      "grad_norm": 2299.00048828125,
      "learning_rate": 7.990881348418554e-06,
      "loss": 124.4509,
      "step": 45330
    },
    {
      "epoch": 0.3750672126401125,
      "grad_norm": 973.8239135742188,
      "learning_rate": 7.989751011266565e-06,
      "loss": 112.5193,
      "step": 45340
    },
    {
      "epoch": 0.37514993588948176,
      "grad_norm": 745.3870239257812,
      "learning_rate": 7.988620436233806e-06,
      "loss": 112.1299,
      "step": 45350
    },
    {
      "epoch": 0.37523265913885095,
      "grad_norm": 2419.01953125,
      "learning_rate": 7.987489623410236e-06,
      "loss": 130.317,
      "step": 45360
    },
    {
      "epoch": 0.3753153823882202,
      "grad_norm": 2003.9739990234375,
      "learning_rate": 7.986358572885828e-06,
      "loss": 118.5811,
      "step": 45370
    },
    {
      "epoch": 0.37539810563758946,
      "grad_norm": 1251.0047607421875,
      "learning_rate": 7.985227284750574e-06,
      "loss": 106.9141,
      "step": 45380
    },
    {
      "epoch": 0.37548082888695866,
      "grad_norm": 879.7947998046875,
      "learning_rate": 7.984095759094485e-06,
      "loss": 93.4843,
      "step": 45390
    },
    {
      "epoch": 0.3755635521363279,
      "grad_norm": 852.9107055664062,
      "learning_rate": 7.982963996007591e-06,
      "loss": 141.075,
      "step": 45400
    },
    {
      "epoch": 0.37564627538569717,
      "grad_norm": 952.68798828125,
      "learning_rate": 7.981831995579943e-06,
      "loss": 91.4658,
      "step": 45410
    },
    {
      "epoch": 0.37572899863506637,
      "grad_norm": 1191.1885986328125,
      "learning_rate": 7.980699757901607e-06,
      "loss": 80.4354,
      "step": 45420
    },
    {
      "epoch": 0.3758117218844356,
      "grad_norm": 813.9268188476562,
      "learning_rate": 7.97956728306267e-06,
      "loss": 90.7331,
      "step": 45430
    },
    {
      "epoch": 0.3758944451338049,
      "grad_norm": 697.1983032226562,
      "learning_rate": 7.97843457115324e-06,
      "loss": 98.3303,
      "step": 45440
    },
    {
      "epoch": 0.3759771683831741,
      "grad_norm": 628.31982421875,
      "learning_rate": 7.97730162226344e-06,
      "loss": 103.7881,
      "step": 45450
    },
    {
      "epoch": 0.37605989163254333,
      "grad_norm": 2688.026123046875,
      "learning_rate": 7.976168436483415e-06,
      "loss": 97.9229,
      "step": 45460
    },
    {
      "epoch": 0.3761426148819126,
      "grad_norm": 813.2411499023438,
      "learning_rate": 7.975035013903326e-06,
      "loss": 94.9314,
      "step": 45470
    },
    {
      "epoch": 0.3762253381312818,
      "grad_norm": 712.8557739257812,
      "learning_rate": 7.973901354613353e-06,
      "loss": 85.9302,
      "step": 45480
    },
    {
      "epoch": 0.37630806138065104,
      "grad_norm": 1530.84033203125,
      "learning_rate": 7.972767458703697e-06,
      "loss": 128.0701,
      "step": 45490
    },
    {
      "epoch": 0.3763907846300203,
      "grad_norm": 1192.7733154296875,
      "learning_rate": 7.971633326264581e-06,
      "loss": 99.9536,
      "step": 45500
    },
    {
      "epoch": 0.3764735078793895,
      "grad_norm": 854.33203125,
      "learning_rate": 7.970498957386237e-06,
      "loss": 121.3529,
      "step": 45510
    },
    {
      "epoch": 0.37655623112875874,
      "grad_norm": 673.5110473632812,
      "learning_rate": 7.969364352158922e-06,
      "loss": 102.1387,
      "step": 45520
    },
    {
      "epoch": 0.376638954378128,
      "grad_norm": 868.6480712890625,
      "learning_rate": 7.968229510672915e-06,
      "loss": 97.5747,
      "step": 45530
    },
    {
      "epoch": 0.3767216776274972,
      "grad_norm": 1399.73779296875,
      "learning_rate": 7.967094433018508e-06,
      "loss": 125.0937,
      "step": 45540
    },
    {
      "epoch": 0.37680440087686645,
      "grad_norm": 1239.8349609375,
      "learning_rate": 7.965959119286013e-06,
      "loss": 113.5951,
      "step": 45550
    },
    {
      "epoch": 0.37688712412623565,
      "grad_norm": 1359.953125,
      "learning_rate": 7.964823569565765e-06,
      "loss": 103.3041,
      "step": 45560
    },
    {
      "epoch": 0.3769698473756049,
      "grad_norm": 720.0075073242188,
      "learning_rate": 7.963687783948111e-06,
      "loss": 81.487,
      "step": 45570
    },
    {
      "epoch": 0.37705257062497416,
      "grad_norm": 1454.417724609375,
      "learning_rate": 7.96255176252342e-06,
      "loss": 83.4532,
      "step": 45580
    },
    {
      "epoch": 0.37713529387434336,
      "grad_norm": 1103.7496337890625,
      "learning_rate": 7.961415505382083e-06,
      "loss": 125.8114,
      "step": 45590
    },
    {
      "epoch": 0.3772180171237126,
      "grad_norm": 948.4056396484375,
      "learning_rate": 7.960279012614508e-06,
      "loss": 90.2628,
      "step": 45600
    },
    {
      "epoch": 0.37730074037308187,
      "grad_norm": 1128.2615966796875,
      "learning_rate": 7.959142284311115e-06,
      "loss": 113.3847,
      "step": 45610
    },
    {
      "epoch": 0.37738346362245107,
      "grad_norm": 921.497802734375,
      "learning_rate": 7.958005320562349e-06,
      "loss": 125.8096,
      "step": 45620
    },
    {
      "epoch": 0.3774661868718203,
      "grad_norm": 1018.8650512695312,
      "learning_rate": 7.95686812145868e-06,
      "loss": 122.7865,
      "step": 45630
    },
    {
      "epoch": 0.3775489101211896,
      "grad_norm": 979.5180053710938,
      "learning_rate": 7.955730687090582e-06,
      "loss": 125.0732,
      "step": 45640
    },
    {
      "epoch": 0.3776316333705588,
      "grad_norm": 978.5775756835938,
      "learning_rate": 7.954593017548557e-06,
      "loss": 87.2099,
      "step": 45650
    },
    {
      "epoch": 0.377714356619928,
      "grad_norm": 978.8362426757812,
      "learning_rate": 7.953455112923127e-06,
      "loss": 134.838,
      "step": 45660
    },
    {
      "epoch": 0.3777970798692973,
      "grad_norm": 961.3007202148438,
      "learning_rate": 7.952316973304828e-06,
      "loss": 109.7581,
      "step": 45670
    },
    {
      "epoch": 0.3778798031186665,
      "grad_norm": 872.6028442382812,
      "learning_rate": 7.951178598784217e-06,
      "loss": 94.058,
      "step": 45680
    },
    {
      "epoch": 0.37796252636803573,
      "grad_norm": 603.7721557617188,
      "learning_rate": 7.950039989451868e-06,
      "loss": 142.0334,
      "step": 45690
    },
    {
      "epoch": 0.378045249617405,
      "grad_norm": 769.1307373046875,
      "learning_rate": 7.948901145398376e-06,
      "loss": 101.5725,
      "step": 45700
    },
    {
      "epoch": 0.3781279728667742,
      "grad_norm": 523.3740844726562,
      "learning_rate": 7.947762066714353e-06,
      "loss": 99.7676,
      "step": 45710
    },
    {
      "epoch": 0.37821069611614344,
      "grad_norm": 569.120361328125,
      "learning_rate": 7.946622753490433e-06,
      "loss": 73.9602,
      "step": 45720
    },
    {
      "epoch": 0.3782934193655127,
      "grad_norm": 850.3909912109375,
      "learning_rate": 7.945483205817262e-06,
      "loss": 99.4327,
      "step": 45730
    },
    {
      "epoch": 0.3783761426148819,
      "grad_norm": 1302.0482177734375,
      "learning_rate": 7.94434342378551e-06,
      "loss": 132.1461,
      "step": 45740
    },
    {
      "epoch": 0.37845886586425115,
      "grad_norm": 698.3201293945312,
      "learning_rate": 7.943203407485864e-06,
      "loss": 69.9039,
      "step": 45750
    },
    {
      "epoch": 0.3785415891136204,
      "grad_norm": 911.8252563476562,
      "learning_rate": 7.942063157009033e-06,
      "loss": 102.4791,
      "step": 45760
    },
    {
      "epoch": 0.3786243123629896,
      "grad_norm": 991.9946899414062,
      "learning_rate": 7.940922672445737e-06,
      "loss": 113.1581,
      "step": 45770
    },
    {
      "epoch": 0.37870703561235886,
      "grad_norm": 1038.773681640625,
      "learning_rate": 7.939781953886722e-06,
      "loss": 128.2211,
      "step": 45780
    },
    {
      "epoch": 0.3787897588617281,
      "grad_norm": 1063.1624755859375,
      "learning_rate": 7.938641001422747e-06,
      "loss": 106.1225,
      "step": 45790
    },
    {
      "epoch": 0.3788724821110973,
      "grad_norm": 897.7556762695312,
      "learning_rate": 7.937499815144597e-06,
      "loss": 102.7701,
      "step": 45800
    },
    {
      "epoch": 0.37895520536046656,
      "grad_norm": 454.84552001953125,
      "learning_rate": 7.936358395143065e-06,
      "loss": 133.0522,
      "step": 45810
    },
    {
      "epoch": 0.3790379286098358,
      "grad_norm": 491.5146484375,
      "learning_rate": 7.935216741508971e-06,
      "loss": 131.0702,
      "step": 45820
    },
    {
      "epoch": 0.379120651859205,
      "grad_norm": 816.953125,
      "learning_rate": 7.934074854333153e-06,
      "loss": 102.6913,
      "step": 45830
    },
    {
      "epoch": 0.37920337510857427,
      "grad_norm": 991.1273193359375,
      "learning_rate": 7.932932733706467e-06,
      "loss": 96.2913,
      "step": 45840
    },
    {
      "epoch": 0.3792860983579435,
      "grad_norm": 2715.989501953125,
      "learning_rate": 7.931790379719781e-06,
      "loss": 145.7241,
      "step": 45850
    },
    {
      "epoch": 0.3793688216073127,
      "grad_norm": 733.826416015625,
      "learning_rate": 7.93064779246399e-06,
      "loss": 79.5076,
      "step": 45860
    },
    {
      "epoch": 0.379451544856682,
      "grad_norm": 1925.5067138671875,
      "learning_rate": 7.929504972030003e-06,
      "loss": 129.3025,
      "step": 45870
    },
    {
      "epoch": 0.37953426810605123,
      "grad_norm": 723.109130859375,
      "learning_rate": 7.928361918508752e-06,
      "loss": 124.5502,
      "step": 45880
    },
    {
      "epoch": 0.37961699135542043,
      "grad_norm": 942.7910766601562,
      "learning_rate": 7.927218631991182e-06,
      "loss": 87.0695,
      "step": 45890
    },
    {
      "epoch": 0.3796997146047897,
      "grad_norm": 1327.8336181640625,
      "learning_rate": 7.92607511256826e-06,
      "loss": 90.3665,
      "step": 45900
    },
    {
      "epoch": 0.37978243785415894,
      "grad_norm": 1204.0654296875,
      "learning_rate": 7.924931360330968e-06,
      "loss": 120.6505,
      "step": 45910
    },
    {
      "epoch": 0.37986516110352814,
      "grad_norm": 643.5101928710938,
      "learning_rate": 7.92378737537031e-06,
      "loss": 84.0929,
      "step": 45920
    },
    {
      "epoch": 0.3799478843528974,
      "grad_norm": 1710.7091064453125,
      "learning_rate": 7.922643157777314e-06,
      "loss": 94.1093,
      "step": 45930
    },
    {
      "epoch": 0.3800306076022666,
      "grad_norm": 1306.8468017578125,
      "learning_rate": 7.921498707643011e-06,
      "loss": 110.1609,
      "step": 45940
    },
    {
      "epoch": 0.38011333085163584,
      "grad_norm": 1266.775146484375,
      "learning_rate": 7.920354025058467e-06,
      "loss": 94.9035,
      "step": 45950
    },
    {
      "epoch": 0.3801960541010051,
      "grad_norm": 1168.3138427734375,
      "learning_rate": 7.919209110114752e-06,
      "loss": 165.2224,
      "step": 45960
    },
    {
      "epoch": 0.3802787773503743,
      "grad_norm": 760.196044921875,
      "learning_rate": 7.918063962902968e-06,
      "loss": 118.3697,
      "step": 45970
    },
    {
      "epoch": 0.38036150059974355,
      "grad_norm": 1145.4422607421875,
      "learning_rate": 7.916918583514227e-06,
      "loss": 102.1873,
      "step": 45980
    },
    {
      "epoch": 0.3804442238491128,
      "grad_norm": 523.3143310546875,
      "learning_rate": 7.91577297203966e-06,
      "loss": 94.5064,
      "step": 45990
    },
    {
      "epoch": 0.380526947098482,
      "grad_norm": 777.9324951171875,
      "learning_rate": 7.91462712857042e-06,
      "loss": 99.4327,
      "step": 46000
    },
    {
      "epoch": 0.38060967034785126,
      "grad_norm": 1038.9837646484375,
      "learning_rate": 7.913481053197673e-06,
      "loss": 120.6933,
      "step": 46010
    },
    {
      "epoch": 0.3806923935972205,
      "grad_norm": 702.442138671875,
      "learning_rate": 7.912334746012613e-06,
      "loss": 141.1689,
      "step": 46020
    },
    {
      "epoch": 0.3807751168465897,
      "grad_norm": 910.6521606445312,
      "learning_rate": 7.911188207106442e-06,
      "loss": 84.091,
      "step": 46030
    },
    {
      "epoch": 0.38085784009595897,
      "grad_norm": 836.0477905273438,
      "learning_rate": 7.910041436570386e-06,
      "loss": 92.7803,
      "step": 46040
    },
    {
      "epoch": 0.3809405633453282,
      "grad_norm": 796.7591552734375,
      "learning_rate": 7.90889443449569e-06,
      "loss": 93.9601,
      "step": 46050
    },
    {
      "epoch": 0.3810232865946974,
      "grad_norm": 1023.9012451171875,
      "learning_rate": 7.90774720097361e-06,
      "loss": 125.3338,
      "step": 46060
    },
    {
      "epoch": 0.3811060098440667,
      "grad_norm": 1169.42822265625,
      "learning_rate": 7.906599736095433e-06,
      "loss": 89.7407,
      "step": 46070
    },
    {
      "epoch": 0.3811887330934359,
      "grad_norm": 924.0548706054688,
      "learning_rate": 7.905452039952453e-06,
      "loss": 86.3304,
      "step": 46080
    },
    {
      "epoch": 0.3812714563428051,
      "grad_norm": 595.2965087890625,
      "learning_rate": 7.904304112635987e-06,
      "loss": 106.9174,
      "step": 46090
    },
    {
      "epoch": 0.3813541795921744,
      "grad_norm": 472.991943359375,
      "learning_rate": 7.903155954237375e-06,
      "loss": 120.8275,
      "step": 46100
    },
    {
      "epoch": 0.38143690284154363,
      "grad_norm": 1009.1423950195312,
      "learning_rate": 7.902007564847967e-06,
      "loss": 105.188,
      "step": 46110
    },
    {
      "epoch": 0.38151962609091283,
      "grad_norm": 851.7573852539062,
      "learning_rate": 7.900858944559133e-06,
      "loss": 111.3603,
      "step": 46120
    },
    {
      "epoch": 0.3816023493402821,
      "grad_norm": 981.0646362304688,
      "learning_rate": 7.899710093462267e-06,
      "loss": 86.6864,
      "step": 46130
    },
    {
      "epoch": 0.38168507258965134,
      "grad_norm": 864.4039306640625,
      "learning_rate": 7.898561011648777e-06,
      "loss": 155.4255,
      "step": 46140
    },
    {
      "epoch": 0.38176779583902054,
      "grad_norm": 1161.7957763671875,
      "learning_rate": 7.89741169921009e-06,
      "loss": 92.5417,
      "step": 46150
    },
    {
      "epoch": 0.3818505190883898,
      "grad_norm": 1559.1328125,
      "learning_rate": 7.896262156237652e-06,
      "loss": 109.0317,
      "step": 46160
    },
    {
      "epoch": 0.38193324233775905,
      "grad_norm": 839.4773559570312,
      "learning_rate": 7.895112382822925e-06,
      "loss": 124.5884,
      "step": 46170
    },
    {
      "epoch": 0.38201596558712825,
      "grad_norm": 1135.74658203125,
      "learning_rate": 7.893962379057393e-06,
      "loss": 115.562,
      "step": 46180
    },
    {
      "epoch": 0.3820986888364975,
      "grad_norm": 979.9491577148438,
      "learning_rate": 7.892812145032557e-06,
      "loss": 118.3164,
      "step": 46190
    },
    {
      "epoch": 0.38218141208586676,
      "grad_norm": 4294.3427734375,
      "learning_rate": 7.891661680839932e-06,
      "loss": 125.0205,
      "step": 46200
    },
    {
      "epoch": 0.38226413533523596,
      "grad_norm": 933.6452026367188,
      "learning_rate": 7.89051098657106e-06,
      "loss": 121.8078,
      "step": 46210
    },
    {
      "epoch": 0.3823468585846052,
      "grad_norm": 527.7791137695312,
      "learning_rate": 7.889360062317495e-06,
      "loss": 96.7531,
      "step": 46220
    },
    {
      "epoch": 0.38242958183397446,
      "grad_norm": 777.3706665039062,
      "learning_rate": 7.888208908170812e-06,
      "loss": 134.1928,
      "step": 46230
    },
    {
      "epoch": 0.38251230508334366,
      "grad_norm": 805.9109497070312,
      "learning_rate": 7.887057524222596e-06,
      "loss": 101.3832,
      "step": 46240
    },
    {
      "epoch": 0.3825950283327129,
      "grad_norm": 1473.904296875,
      "learning_rate": 7.885905910564466e-06,
      "loss": 87.6336,
      "step": 46250
    },
    {
      "epoch": 0.38267775158208217,
      "grad_norm": 784.258056640625,
      "learning_rate": 7.884754067288047e-06,
      "loss": 94.394,
      "step": 46260
    },
    {
      "epoch": 0.38276047483145137,
      "grad_norm": 1091.794677734375,
      "learning_rate": 7.883601994484986e-06,
      "loss": 106.1401,
      "step": 46270
    },
    {
      "epoch": 0.3828431980808206,
      "grad_norm": 1147.5867919921875,
      "learning_rate": 7.882449692246948e-06,
      "loss": 104.8446,
      "step": 46280
    },
    {
      "epoch": 0.3829259213301898,
      "grad_norm": 1361.256591796875,
      "learning_rate": 7.881297160665616e-06,
      "loss": 97.2336,
      "step": 46290
    },
    {
      "epoch": 0.3830086445795591,
      "grad_norm": 1118.20556640625,
      "learning_rate": 7.880144399832693e-06,
      "loss": 109.2898,
      "step": 46300
    },
    {
      "epoch": 0.38309136782892833,
      "grad_norm": 1180.1666259765625,
      "learning_rate": 7.878991409839897e-06,
      "loss": 107.2897,
      "step": 46310
    },
    {
      "epoch": 0.38317409107829753,
      "grad_norm": 976.9520874023438,
      "learning_rate": 7.87783819077897e-06,
      "loss": 127.2354,
      "step": 46320
    },
    {
      "epoch": 0.3832568143276668,
      "grad_norm": 569.0108032226562,
      "learning_rate": 7.876684742741665e-06,
      "loss": 89.219,
      "step": 46330
    },
    {
      "epoch": 0.38333953757703604,
      "grad_norm": 1490.4720458984375,
      "learning_rate": 7.875531065819755e-06,
      "loss": 88.2369,
      "step": 46340
    },
    {
      "epoch": 0.38342226082640524,
      "grad_norm": 1101.564697265625,
      "learning_rate": 7.874377160105037e-06,
      "loss": 104.7651,
      "step": 46350
    },
    {
      "epoch": 0.3835049840757745,
      "grad_norm": 1799.0706787109375,
      "learning_rate": 7.873223025689319e-06,
      "loss": 138.4782,
      "step": 46360
    },
    {
      "epoch": 0.38358770732514375,
      "grad_norm": 1268.860595703125,
      "learning_rate": 7.872068662664432e-06,
      "loss": 91.4783,
      "step": 46370
    },
    {
      "epoch": 0.38367043057451294,
      "grad_norm": 1431.305419921875,
      "learning_rate": 7.870914071122222e-06,
      "loss": 100.8851,
      "step": 46380
    },
    {
      "epoch": 0.3837531538238822,
      "grad_norm": 1034.6007080078125,
      "learning_rate": 7.869759251154554e-06,
      "loss": 94.5458,
      "step": 46390
    },
    {
      "epoch": 0.38383587707325145,
      "grad_norm": 1329.562255859375,
      "learning_rate": 7.868604202853314e-06,
      "loss": 95.9945,
      "step": 46400
    },
    {
      "epoch": 0.38391860032262065,
      "grad_norm": 755.7239379882812,
      "learning_rate": 7.867448926310403e-06,
      "loss": 106.9036,
      "step": 46410
    },
    {
      "epoch": 0.3840013235719899,
      "grad_norm": 1034.0926513671875,
      "learning_rate": 7.866293421617741e-06,
      "loss": 94.1212,
      "step": 46420
    },
    {
      "epoch": 0.38408404682135916,
      "grad_norm": 1169.0289306640625,
      "learning_rate": 7.865137688867264e-06,
      "loss": 76.2746,
      "step": 46430
    },
    {
      "epoch": 0.38416677007072836,
      "grad_norm": 1374.6610107421875,
      "learning_rate": 7.86398172815093e-06,
      "loss": 141.711,
      "step": 46440
    },
    {
      "epoch": 0.3842494933200976,
      "grad_norm": 715.2680053710938,
      "learning_rate": 7.862825539560716e-06,
      "loss": 127.8841,
      "step": 46450
    },
    {
      "epoch": 0.38433221656946687,
      "grad_norm": 1387.7430419921875,
      "learning_rate": 7.861669123188613e-06,
      "loss": 124.0174,
      "step": 46460
    },
    {
      "epoch": 0.38441493981883607,
      "grad_norm": 826.4485473632812,
      "learning_rate": 7.86051247912663e-06,
      "loss": 110.8221,
      "step": 46470
    },
    {
      "epoch": 0.3844976630682053,
      "grad_norm": 807.095703125,
      "learning_rate": 7.859355607466797e-06,
      "loss": 87.1465,
      "step": 46480
    },
    {
      "epoch": 0.3845803863175746,
      "grad_norm": 1193.6893310546875,
      "learning_rate": 7.858198508301161e-06,
      "loss": 133.6182,
      "step": 46490
    },
    {
      "epoch": 0.3846631095669438,
      "grad_norm": 527.4124755859375,
      "learning_rate": 7.857041181721788e-06,
      "loss": 97.8456,
      "step": 46500
    },
    {
      "epoch": 0.384745832816313,
      "grad_norm": 862.8912963867188,
      "learning_rate": 7.855883627820757e-06,
      "loss": 134.22,
      "step": 46510
    },
    {
      "epoch": 0.3848285560656823,
      "grad_norm": 548.6507568359375,
      "learning_rate": 7.854725846690175e-06,
      "loss": 83.1777,
      "step": 46520
    },
    {
      "epoch": 0.3849112793150515,
      "grad_norm": 902.584228515625,
      "learning_rate": 7.85356783842216e-06,
      "loss": 106.9863,
      "step": 46530
    },
    {
      "epoch": 0.38499400256442073,
      "grad_norm": 535.5631103515625,
      "learning_rate": 7.852409603108845e-06,
      "loss": 113.9048,
      "step": 46540
    },
    {
      "epoch": 0.38507672581379,
      "grad_norm": 938.954345703125,
      "learning_rate": 7.85125114084239e-06,
      "loss": 105.1793,
      "step": 46550
    },
    {
      "epoch": 0.3851594490631592,
      "grad_norm": 1107.5294189453125,
      "learning_rate": 7.850092451714967e-06,
      "loss": 115.7462,
      "step": 46560
    },
    {
      "epoch": 0.38524217231252844,
      "grad_norm": 1338.5953369140625,
      "learning_rate": 7.84893353581877e-06,
      "loss": 93.1394,
      "step": 46570
    },
    {
      "epoch": 0.3853248955618977,
      "grad_norm": 1115.7308349609375,
      "learning_rate": 7.847774393246005e-06,
      "loss": 117.0743,
      "step": 46580
    },
    {
      "epoch": 0.3854076188112669,
      "grad_norm": 712.0968017578125,
      "learning_rate": 7.8466150240889e-06,
      "loss": 92.4161,
      "step": 46590
    },
    {
      "epoch": 0.38549034206063615,
      "grad_norm": 1130.7666015625,
      "learning_rate": 7.845455428439703e-06,
      "loss": 90.2969,
      "step": 46600
    },
    {
      "epoch": 0.3855730653100054,
      "grad_norm": 694.2098388671875,
      "learning_rate": 7.844295606390675e-06,
      "loss": 106.4844,
      "step": 46610
    },
    {
      "epoch": 0.3856557885593746,
      "grad_norm": 758.242431640625,
      "learning_rate": 7.843135558034101e-06,
      "loss": 85.5782,
      "step": 46620
    },
    {
      "epoch": 0.38573851180874386,
      "grad_norm": 847.0222778320312,
      "learning_rate": 7.841975283462278e-06,
      "loss": 83.5226,
      "step": 46630
    },
    {
      "epoch": 0.3858212350581131,
      "grad_norm": 828.2359619140625,
      "learning_rate": 7.840814782767525e-06,
      "loss": 74.5017,
      "step": 46640
    },
    {
      "epoch": 0.3859039583074823,
      "grad_norm": 575.1474609375,
      "learning_rate": 7.839654056042176e-06,
      "loss": 84.7638,
      "step": 46650
    },
    {
      "epoch": 0.38598668155685156,
      "grad_norm": 961.8069458007812,
      "learning_rate": 7.838493103378588e-06,
      "loss": 105.9289,
      "step": 46660
    },
    {
      "epoch": 0.38606940480622076,
      "grad_norm": 323.90765380859375,
      "learning_rate": 7.83733192486913e-06,
      "loss": 134.9893,
      "step": 46670
    },
    {
      "epoch": 0.38615212805559,
      "grad_norm": 906.62060546875,
      "learning_rate": 7.836170520606191e-06,
      "loss": 132.9467,
      "step": 46680
    },
    {
      "epoch": 0.38623485130495927,
      "grad_norm": 1547.6925048828125,
      "learning_rate": 7.83500889068218e-06,
      "loss": 145.7491,
      "step": 46690
    },
    {
      "epoch": 0.38631757455432847,
      "grad_norm": 1164.046142578125,
      "learning_rate": 7.833847035189524e-06,
      "loss": 110.9907,
      "step": 46700
    },
    {
      "epoch": 0.3864002978036977,
      "grad_norm": 1105.657470703125,
      "learning_rate": 7.832684954220664e-06,
      "loss": 124.7499,
      "step": 46710
    },
    {
      "epoch": 0.386483021053067,
      "grad_norm": 508.67803955078125,
      "learning_rate": 7.831522647868064e-06,
      "loss": 95.8649,
      "step": 46720
    },
    {
      "epoch": 0.3865657443024362,
      "grad_norm": 674.171875,
      "learning_rate": 7.8303601162242e-06,
      "loss": 99.7405,
      "step": 46730
    },
    {
      "epoch": 0.38664846755180543,
      "grad_norm": 1009.1093139648438,
      "learning_rate": 7.829197359381571e-06,
      "loss": 111.1325,
      "step": 46740
    },
    {
      "epoch": 0.3867311908011747,
      "grad_norm": 1165.2833251953125,
      "learning_rate": 7.828034377432694e-06,
      "loss": 111.5451,
      "step": 46750
    },
    {
      "epoch": 0.3868139140505439,
      "grad_norm": 542.6980590820312,
      "learning_rate": 7.826871170470099e-06,
      "loss": 93.1879,
      "step": 46760
    },
    {
      "epoch": 0.38689663729991314,
      "grad_norm": 1133.044921875,
      "learning_rate": 7.82570773858634e-06,
      "loss": 90.8988,
      "step": 46770
    },
    {
      "epoch": 0.3869793605492824,
      "grad_norm": 684.3965454101562,
      "learning_rate": 7.824544081873984e-06,
      "loss": 79.4068,
      "step": 46780
    },
    {
      "epoch": 0.3870620837986516,
      "grad_norm": 1281.01513671875,
      "learning_rate": 7.823380200425618e-06,
      "loss": 132.8948,
      "step": 46790
    },
    {
      "epoch": 0.38714480704802084,
      "grad_norm": 1106.5760498046875,
      "learning_rate": 7.822216094333847e-06,
      "loss": 109.8843,
      "step": 46800
    },
    {
      "epoch": 0.3872275302973901,
      "grad_norm": 736.9039306640625,
      "learning_rate": 7.821051763691293e-06,
      "loss": 95.4937,
      "step": 46810
    },
    {
      "epoch": 0.3873102535467593,
      "grad_norm": 983.0117797851562,
      "learning_rate": 7.819887208590597e-06,
      "loss": 106.8212,
      "step": 46820
    },
    {
      "epoch": 0.38739297679612855,
      "grad_norm": 900.6527099609375,
      "learning_rate": 7.818722429124418e-06,
      "loss": 97.8815,
      "step": 46830
    },
    {
      "epoch": 0.3874757000454978,
      "grad_norm": 850.8548583984375,
      "learning_rate": 7.817557425385433e-06,
      "loss": 81.9495,
      "step": 46840
    },
    {
      "epoch": 0.387558423294867,
      "grad_norm": 855.3139038085938,
      "learning_rate": 7.816392197466333e-06,
      "loss": 89.8393,
      "step": 46850
    },
    {
      "epoch": 0.38764114654423626,
      "grad_norm": 587.1494750976562,
      "learning_rate": 7.815226745459831e-06,
      "loss": 106.3622,
      "step": 46860
    },
    {
      "epoch": 0.3877238697936055,
      "grad_norm": 3245.526123046875,
      "learning_rate": 7.814061069458657e-06,
      "loss": 111.3302,
      "step": 46870
    },
    {
      "epoch": 0.3878065930429747,
      "grad_norm": 547.8377685546875,
      "learning_rate": 7.81289516955556e-06,
      "loss": 78.6426,
      "step": 46880
    },
    {
      "epoch": 0.38788931629234397,
      "grad_norm": 1064.7379150390625,
      "learning_rate": 7.811729045843303e-06,
      "loss": 90.7304,
      "step": 46890
    },
    {
      "epoch": 0.3879720395417132,
      "grad_norm": 616.6322631835938,
      "learning_rate": 7.81056269841467e-06,
      "loss": 107.5491,
      "step": 46900
    },
    {
      "epoch": 0.3880547627910824,
      "grad_norm": 1086.90185546875,
      "learning_rate": 7.80939612736246e-06,
      "loss": 112.952,
      "step": 46910
    },
    {
      "epoch": 0.3881374860404517,
      "grad_norm": 762.38916015625,
      "learning_rate": 7.808229332779496e-06,
      "loss": 93.649,
      "step": 46920
    },
    {
      "epoch": 0.38822020928982093,
      "grad_norm": 1226.2991943359375,
      "learning_rate": 7.807062314758612e-06,
      "loss": 122.8837,
      "step": 46930
    },
    {
      "epoch": 0.3883029325391901,
      "grad_norm": 980.0850219726562,
      "learning_rate": 7.80589507339266e-06,
      "loss": 118.3508,
      "step": 46940
    },
    {
      "epoch": 0.3883856557885594,
      "grad_norm": 899.0253295898438,
      "learning_rate": 7.804727608774516e-06,
      "loss": 95.3277,
      "step": 46950
    },
    {
      "epoch": 0.38846837903792864,
      "grad_norm": 992.0533447265625,
      "learning_rate": 7.803559920997067e-06,
      "loss": 100.9585,
      "step": 46960
    },
    {
      "epoch": 0.38855110228729783,
      "grad_norm": 1150.515869140625,
      "learning_rate": 7.802392010153223e-06,
      "loss": 106.7416,
      "step": 46970
    },
    {
      "epoch": 0.3886338255366671,
      "grad_norm": 392.8724670410156,
      "learning_rate": 7.801223876335907e-06,
      "loss": 81.8418,
      "step": 46980
    },
    {
      "epoch": 0.38871654878603634,
      "grad_norm": 958.8024291992188,
      "learning_rate": 7.800055519638064e-06,
      "loss": 104.2003,
      "step": 46990
    },
    {
      "epoch": 0.38879927203540554,
      "grad_norm": 1033.255859375,
      "learning_rate": 7.798886940152654e-06,
      "loss": 111.1013,
      "step": 47000
    },
    {
      "epoch": 0.3888819952847748,
      "grad_norm": 670.1970825195312,
      "learning_rate": 7.797718137972654e-06,
      "loss": 108.9194,
      "step": 47010
    },
    {
      "epoch": 0.388964718534144,
      "grad_norm": 927.21630859375,
      "learning_rate": 7.79654911319106e-06,
      "loss": 107.1055,
      "step": 47020
    },
    {
      "epoch": 0.38904744178351325,
      "grad_norm": 1264.6552734375,
      "learning_rate": 7.795379865900892e-06,
      "loss": 131.0568,
      "step": 47030
    },
    {
      "epoch": 0.3891301650328825,
      "grad_norm": 1989.842529296875,
      "learning_rate": 7.794210396195175e-06,
      "loss": 94.6299,
      "step": 47040
    },
    {
      "epoch": 0.3892128882822517,
      "grad_norm": 530.0291748046875,
      "learning_rate": 7.79304070416696e-06,
      "loss": 121.8401,
      "step": 47050
    },
    {
      "epoch": 0.38929561153162096,
      "grad_norm": 936.9892578125,
      "learning_rate": 7.791870789909315e-06,
      "loss": 89.362,
      "step": 47060
    },
    {
      "epoch": 0.3893783347809902,
      "grad_norm": 2093.591796875,
      "learning_rate": 7.790700653515324e-06,
      "loss": 90.1807,
      "step": 47070
    },
    {
      "epoch": 0.3894610580303594,
      "grad_norm": 549.9674682617188,
      "learning_rate": 7.789530295078089e-06,
      "loss": 100.1859,
      "step": 47080
    },
    {
      "epoch": 0.38954378127972866,
      "grad_norm": 882.615966796875,
      "learning_rate": 7.788359714690732e-06,
      "loss": 117.6557,
      "step": 47090
    },
    {
      "epoch": 0.3896265045290979,
      "grad_norm": 798.0199584960938,
      "learning_rate": 7.787188912446389e-06,
      "loss": 127.0244,
      "step": 47100
    },
    {
      "epoch": 0.3897092277784671,
      "grad_norm": 776.5193481445312,
      "learning_rate": 7.786017888438214e-06,
      "loss": 126.0215,
      "step": 47110
    },
    {
      "epoch": 0.38979195102783637,
      "grad_norm": 654.182373046875,
      "learning_rate": 7.784846642759383e-06,
      "loss": 82.0503,
      "step": 47120
    },
    {
      "epoch": 0.3898746742772056,
      "grad_norm": 822.113525390625,
      "learning_rate": 7.783675175503087e-06,
      "loss": 89.8243,
      "step": 47130
    },
    {
      "epoch": 0.3899573975265748,
      "grad_norm": 865.35546875,
      "learning_rate": 7.78250348676253e-06,
      "loss": 98.9063,
      "step": 47140
    },
    {
      "epoch": 0.3900401207759441,
      "grad_norm": 1154.8814697265625,
      "learning_rate": 7.781331576630941e-06,
      "loss": 146.3992,
      "step": 47150
    },
    {
      "epoch": 0.39012284402531333,
      "grad_norm": 1378.4947509765625,
      "learning_rate": 7.780159445201562e-06,
      "loss": 130.5315,
      "step": 47160
    },
    {
      "epoch": 0.39020556727468253,
      "grad_norm": 1242.33251953125,
      "learning_rate": 7.778987092567658e-06,
      "loss": 119.4341,
      "step": 47170
    },
    {
      "epoch": 0.3902882905240518,
      "grad_norm": 797.1754760742188,
      "learning_rate": 7.777814518822504e-06,
      "loss": 67.0835,
      "step": 47180
    },
    {
      "epoch": 0.39037101377342104,
      "grad_norm": 607.6759033203125,
      "learning_rate": 7.776641724059398e-06,
      "loss": 93.4001,
      "step": 47190
    },
    {
      "epoch": 0.39045373702279024,
      "grad_norm": 819.6945190429688,
      "learning_rate": 7.77546870837165e-06,
      "loss": 74.3229,
      "step": 47200
    },
    {
      "epoch": 0.3905364602721595,
      "grad_norm": 903.440185546875,
      "learning_rate": 7.774295471852596e-06,
      "loss": 108.8114,
      "step": 47210
    },
    {
      "epoch": 0.39061918352152875,
      "grad_norm": 785.1321411132812,
      "learning_rate": 7.773122014595584e-06,
      "loss": 169.5685,
      "step": 47220
    },
    {
      "epoch": 0.39070190677089794,
      "grad_norm": 1930.775146484375,
      "learning_rate": 7.771948336693983e-06,
      "loss": 108.4483,
      "step": 47230
    },
    {
      "epoch": 0.3907846300202672,
      "grad_norm": 515.0844116210938,
      "learning_rate": 7.770774438241168e-06,
      "loss": 67.2212,
      "step": 47240
    },
    {
      "epoch": 0.39086735326963645,
      "grad_norm": 1046.875,
      "learning_rate": 7.769600319330553e-06,
      "loss": 122.4751,
      "step": 47250
    },
    {
      "epoch": 0.39095007651900565,
      "grad_norm": 516.0284423828125,
      "learning_rate": 7.768425980055548e-06,
      "loss": 87.5364,
      "step": 47260
    },
    {
      "epoch": 0.3910327997683749,
      "grad_norm": 1055.4755859375,
      "learning_rate": 7.767251420509593e-06,
      "loss": 127.459,
      "step": 47270
    },
    {
      "epoch": 0.39111552301774416,
      "grad_norm": 1103.45361328125,
      "learning_rate": 7.766076640786145e-06,
      "loss": 91.529,
      "step": 47280
    },
    {
      "epoch": 0.39119824626711336,
      "grad_norm": 751.03759765625,
      "learning_rate": 7.764901640978671e-06,
      "loss": 97.0965,
      "step": 47290
    },
    {
      "epoch": 0.3912809695164826,
      "grad_norm": 849.8720703125,
      "learning_rate": 7.763726421180664e-06,
      "loss": 100.8384,
      "step": 47300
    },
    {
      "epoch": 0.39136369276585187,
      "grad_norm": 1227.1131591796875,
      "learning_rate": 7.762550981485629e-06,
      "loss": 85.1875,
      "step": 47310
    },
    {
      "epoch": 0.39144641601522107,
      "grad_norm": 1593.5885009765625,
      "learning_rate": 7.76137532198709e-06,
      "loss": 117.0101,
      "step": 47320
    },
    {
      "epoch": 0.3915291392645903,
      "grad_norm": 901.8778686523438,
      "learning_rate": 7.76019944277859e-06,
      "loss": 132.7024,
      "step": 47330
    },
    {
      "epoch": 0.3916118625139596,
      "grad_norm": 851.2544555664062,
      "learning_rate": 7.759023343953689e-06,
      "loss": 89.3048,
      "step": 47340
    },
    {
      "epoch": 0.3916945857633288,
      "grad_norm": 605.6761474609375,
      "learning_rate": 7.757847025605963e-06,
      "loss": 103.3425,
      "step": 47350
    },
    {
      "epoch": 0.391777309012698,
      "grad_norm": 1666.5958251953125,
      "learning_rate": 7.756670487829005e-06,
      "loss": 112.8522,
      "step": 47360
    },
    {
      "epoch": 0.3918600322620673,
      "grad_norm": 928.7994384765625,
      "learning_rate": 7.755493730716428e-06,
      "loss": 139.6545,
      "step": 47370
    },
    {
      "epoch": 0.3919427555114365,
      "grad_norm": 1118.8592529296875,
      "learning_rate": 7.75431675436186e-06,
      "loss": 101.6983,
      "step": 47380
    },
    {
      "epoch": 0.39202547876080573,
      "grad_norm": 753.40185546875,
      "learning_rate": 7.753139558858949e-06,
      "loss": 111.5847,
      "step": 47390
    },
    {
      "epoch": 0.39210820201017493,
      "grad_norm": 1253.97509765625,
      "learning_rate": 7.751962144301359e-06,
      "loss": 101.4553,
      "step": 47400
    },
    {
      "epoch": 0.3921909252595442,
      "grad_norm": 1033.0867919921875,
      "learning_rate": 7.75078451078277e-06,
      "loss": 92.0257,
      "step": 47410
    },
    {
      "epoch": 0.39227364850891344,
      "grad_norm": 747.4989624023438,
      "learning_rate": 7.749606658396883e-06,
      "loss": 100.6043,
      "step": 47420
    },
    {
      "epoch": 0.39235637175828264,
      "grad_norm": 822.0686645507812,
      "learning_rate": 7.748428587237412e-06,
      "loss": 80.0977,
      "step": 47430
    },
    {
      "epoch": 0.3924390950076519,
      "grad_norm": 1061.1436767578125,
      "learning_rate": 7.747250297398092e-06,
      "loss": 120.7229,
      "step": 47440
    },
    {
      "epoch": 0.39252181825702115,
      "grad_norm": 497.0961608886719,
      "learning_rate": 7.746071788972675e-06,
      "loss": 92.1028,
      "step": 47450
    },
    {
      "epoch": 0.39260454150639035,
      "grad_norm": 747.8739624023438,
      "learning_rate": 7.744893062054928e-06,
      "loss": 84.091,
      "step": 47460
    },
    {
      "epoch": 0.3926872647557596,
      "grad_norm": 1407.6231689453125,
      "learning_rate": 7.743714116738636e-06,
      "loss": 96.2375,
      "step": 47470
    },
    {
      "epoch": 0.39276998800512886,
      "grad_norm": 1389.636962890625,
      "learning_rate": 7.742534953117607e-06,
      "loss": 83.5301,
      "step": 47480
    },
    {
      "epoch": 0.39285271125449805,
      "grad_norm": 827.8368530273438,
      "learning_rate": 7.741355571285656e-06,
      "loss": 98.6002,
      "step": 47490
    },
    {
      "epoch": 0.3929354345038673,
      "grad_norm": 1032.63330078125,
      "learning_rate": 7.740175971336624e-06,
      "loss": 111.3695,
      "step": 47500
    },
    {
      "epoch": 0.39301815775323656,
      "grad_norm": 938.416015625,
      "learning_rate": 7.738996153364364e-06,
      "loss": 128.9613,
      "step": 47510
    },
    {
      "epoch": 0.39310088100260576,
      "grad_norm": 960.8037719726562,
      "learning_rate": 7.737816117462752e-06,
      "loss": 96.7483,
      "step": 47520
    },
    {
      "epoch": 0.393183604251975,
      "grad_norm": 880.8233642578125,
      "learning_rate": 7.736635863725677e-06,
      "loss": 112.1702,
      "step": 47530
    },
    {
      "epoch": 0.39326632750134427,
      "grad_norm": 643.0890502929688,
      "learning_rate": 7.735455392247044e-06,
      "loss": 172.7514,
      "step": 47540
    },
    {
      "epoch": 0.39334905075071347,
      "grad_norm": 481.9482421875,
      "learning_rate": 7.73427470312078e-06,
      "loss": 98.0854,
      "step": 47550
    },
    {
      "epoch": 0.3934317740000827,
      "grad_norm": 7659.08837890625,
      "learning_rate": 7.733093796440828e-06,
      "loss": 94.3791,
      "step": 47560
    },
    {
      "epoch": 0.393514497249452,
      "grad_norm": 611.7446899414062,
      "learning_rate": 7.731912672301145e-06,
      "loss": 121.1342,
      "step": 47570
    },
    {
      "epoch": 0.3935972204988212,
      "grad_norm": 1074.301513671875,
      "learning_rate": 7.730731330795707e-06,
      "loss": 107.5944,
      "step": 47580
    },
    {
      "epoch": 0.39367994374819043,
      "grad_norm": 1130.7880859375,
      "learning_rate": 7.72954977201851e-06,
      "loss": 88.7905,
      "step": 47590
    },
    {
      "epoch": 0.3937626669975597,
      "grad_norm": 427.6048278808594,
      "learning_rate": 7.728367996063566e-06,
      "loss": 95.4467,
      "step": 47600
    },
    {
      "epoch": 0.3938453902469289,
      "grad_norm": 1087.6966552734375,
      "learning_rate": 7.727186003024902e-06,
      "loss": 116.7486,
      "step": 47610
    },
    {
      "epoch": 0.39392811349629814,
      "grad_norm": 677.6015014648438,
      "learning_rate": 7.726003792996562e-06,
      "loss": 112.2149,
      "step": 47620
    },
    {
      "epoch": 0.3940108367456674,
      "grad_norm": 1281.115478515625,
      "learning_rate": 7.724821366072612e-06,
      "loss": 112.6385,
      "step": 47630
    },
    {
      "epoch": 0.3940935599950366,
      "grad_norm": 1624.0828857421875,
      "learning_rate": 7.723638722347132e-06,
      "loss": 114.2262,
      "step": 47640
    },
    {
      "epoch": 0.39417628324440585,
      "grad_norm": 835.2601318359375,
      "learning_rate": 7.722455861914218e-06,
      "loss": 87.17,
      "step": 47650
    },
    {
      "epoch": 0.3942590064937751,
      "grad_norm": 721.5057983398438,
      "learning_rate": 7.721272784867983e-06,
      "loss": 112.5632,
      "step": 47660
    },
    {
      "epoch": 0.3943417297431443,
      "grad_norm": 909.7614135742188,
      "learning_rate": 7.720089491302565e-06,
      "loss": 105.2768,
      "step": 47670
    },
    {
      "epoch": 0.39442445299251355,
      "grad_norm": 641.9271240234375,
      "learning_rate": 7.718905981312108e-06,
      "loss": 93.7019,
      "step": 47680
    },
    {
      "epoch": 0.3945071762418828,
      "grad_norm": 585.551025390625,
      "learning_rate": 7.71772225499078e-06,
      "loss": 80.3911,
      "step": 47690
    },
    {
      "epoch": 0.394589899491252,
      "grad_norm": 783.8123168945312,
      "learning_rate": 7.716538312432767e-06,
      "loss": 115.101,
      "step": 47700
    },
    {
      "epoch": 0.39467262274062126,
      "grad_norm": 644.3832397460938,
      "learning_rate": 7.715354153732265e-06,
      "loss": 99.6165,
      "step": 47710
    },
    {
      "epoch": 0.3947553459899905,
      "grad_norm": 1999.86572265625,
      "learning_rate": 7.714169778983496e-06,
      "loss": 117.3061,
      "step": 47720
    },
    {
      "epoch": 0.3948380692393597,
      "grad_norm": 1040.9405517578125,
      "learning_rate": 7.712985188280694e-06,
      "loss": 101.7906,
      "step": 47730
    },
    {
      "epoch": 0.39492079248872897,
      "grad_norm": 1040.4888916015625,
      "learning_rate": 7.711800381718111e-06,
      "loss": 110.293,
      "step": 47740
    },
    {
      "epoch": 0.39500351573809817,
      "grad_norm": 1044.0440673828125,
      "learning_rate": 7.710615359390018e-06,
      "loss": 94.9559,
      "step": 47750
    },
    {
      "epoch": 0.3950862389874674,
      "grad_norm": 749.12939453125,
      "learning_rate": 7.7094301213907e-06,
      "loss": 92.2077,
      "step": 47760
    },
    {
      "epoch": 0.3951689622368367,
      "grad_norm": 1238.37158203125,
      "learning_rate": 7.708244667814463e-06,
      "loss": 100.6183,
      "step": 47770
    },
    {
      "epoch": 0.3952516854862059,
      "grad_norm": 821.638916015625,
      "learning_rate": 7.707058998755626e-06,
      "loss": 109.0208,
      "step": 47780
    },
    {
      "epoch": 0.3953344087355751,
      "grad_norm": 767.8424072265625,
      "learning_rate": 7.705873114308529e-06,
      "loss": 85.0486,
      "step": 47790
    },
    {
      "epoch": 0.3954171319849444,
      "grad_norm": 756.989501953125,
      "learning_rate": 7.704687014567524e-06,
      "loss": 89.8211,
      "step": 47800
    },
    {
      "epoch": 0.3954998552343136,
      "grad_norm": 1021.226806640625,
      "learning_rate": 7.703500699626988e-06,
      "loss": 90.8781,
      "step": 47810
    },
    {
      "epoch": 0.39558257848368283,
      "grad_norm": 749.4614868164062,
      "learning_rate": 7.702314169581311e-06,
      "loss": 103.451,
      "step": 47820
    },
    {
      "epoch": 0.3956653017330521,
      "grad_norm": 954.7346801757812,
      "learning_rate": 7.701127424524894e-06,
      "loss": 97.5412,
      "step": 47830
    },
    {
      "epoch": 0.3957480249824213,
      "grad_norm": 704.9465942382812,
      "learning_rate": 7.699940464552166e-06,
      "loss": 128.5169,
      "step": 47840
    },
    {
      "epoch": 0.39583074823179054,
      "grad_norm": 926.2518310546875,
      "learning_rate": 7.698753289757565e-06,
      "loss": 103.6893,
      "step": 47850
    },
    {
      "epoch": 0.3959134714811598,
      "grad_norm": 903.693115234375,
      "learning_rate": 7.69756590023555e-06,
      "loss": 117.7347,
      "step": 47860
    },
    {
      "epoch": 0.395996194730529,
      "grad_norm": 576.5755004882812,
      "learning_rate": 7.696378296080598e-06,
      "loss": 77.2486,
      "step": 47870
    },
    {
      "epoch": 0.39607891797989825,
      "grad_norm": 1016.8932495117188,
      "learning_rate": 7.6951904773872e-06,
      "loss": 102.5683,
      "step": 47880
    },
    {
      "epoch": 0.3961616412292675,
      "grad_norm": 649.89013671875,
      "learning_rate": 7.694002444249863e-06,
      "loss": 101.7743,
      "step": 47890
    },
    {
      "epoch": 0.3962443644786367,
      "grad_norm": 2596.306396484375,
      "learning_rate": 7.692814196763118e-06,
      "loss": 125.352,
      "step": 47900
    },
    {
      "epoch": 0.39632708772800596,
      "grad_norm": 1052.6181640625,
      "learning_rate": 7.691625735021505e-06,
      "loss": 109.4487,
      "step": 47910
    },
    {
      "epoch": 0.3964098109773752,
      "grad_norm": 909.6773071289062,
      "learning_rate": 7.690437059119584e-06,
      "loss": 132.6711,
      "step": 47920
    },
    {
      "epoch": 0.3964925342267444,
      "grad_norm": 468.37310791015625,
      "learning_rate": 7.689248169151935e-06,
      "loss": 89.8137,
      "step": 47930
    },
    {
      "epoch": 0.39657525747611366,
      "grad_norm": 1061.165771484375,
      "learning_rate": 7.68805906521315e-06,
      "loss": 117.4328,
      "step": 47940
    },
    {
      "epoch": 0.3966579807254829,
      "grad_norm": 757.1315307617188,
      "learning_rate": 7.686869747397843e-06,
      "loss": 101.1938,
      "step": 47950
    },
    {
      "epoch": 0.3967407039748521,
      "grad_norm": 811.54052734375,
      "learning_rate": 7.685680215800639e-06,
      "loss": 98.3037,
      "step": 47960
    },
    {
      "epoch": 0.39682342722422137,
      "grad_norm": 880.4942626953125,
      "learning_rate": 7.684490470516185e-06,
      "loss": 105.742,
      "step": 47970
    },
    {
      "epoch": 0.3969061504735906,
      "grad_norm": 3485.52685546875,
      "learning_rate": 7.683300511639149e-06,
      "loss": 121.3876,
      "step": 47980
    },
    {
      "epoch": 0.3969888737229598,
      "grad_norm": 794.6721801757812,
      "learning_rate": 7.682110339264203e-06,
      "loss": 99.216,
      "step": 47990
    },
    {
      "epoch": 0.3970715969723291,
      "grad_norm": 1388.62255859375,
      "learning_rate": 7.680919953486047e-06,
      "loss": 114.4334,
      "step": 48000
    },
    {
      "epoch": 0.39715432022169833,
      "grad_norm": 617.267578125,
      "learning_rate": 7.679729354399395e-06,
      "loss": 96.2605,
      "step": 48010
    },
    {
      "epoch": 0.39723704347106753,
      "grad_norm": 577.44189453125,
      "learning_rate": 7.678538542098974e-06,
      "loss": 106.764,
      "step": 48020
    },
    {
      "epoch": 0.3973197667204368,
      "grad_norm": 922.736083984375,
      "learning_rate": 7.677347516679536e-06,
      "loss": 94.4455,
      "step": 48030
    },
    {
      "epoch": 0.39740248996980604,
      "grad_norm": 1115.3370361328125,
      "learning_rate": 7.676156278235845e-06,
      "loss": 124.8899,
      "step": 48040
    },
    {
      "epoch": 0.39748521321917524,
      "grad_norm": 956.1436157226562,
      "learning_rate": 7.674964826862679e-06,
      "loss": 114.437,
      "step": 48050
    },
    {
      "epoch": 0.3975679364685445,
      "grad_norm": 1833.231201171875,
      "learning_rate": 7.673773162654836e-06,
      "loss": 127.1646,
      "step": 48060
    },
    {
      "epoch": 0.39765065971791375,
      "grad_norm": 424.2684631347656,
      "learning_rate": 7.672581285707135e-06,
      "loss": 88.6944,
      "step": 48070
    },
    {
      "epoch": 0.39773338296728294,
      "grad_norm": 1856.982421875,
      "learning_rate": 7.67138919611441e-06,
      "loss": 154.6424,
      "step": 48080
    },
    {
      "epoch": 0.3978161062166522,
      "grad_norm": 907.585205078125,
      "learning_rate": 7.670196893971502e-06,
      "loss": 121.6254,
      "step": 48090
    },
    {
      "epoch": 0.3978988294660214,
      "grad_norm": 1184.2100830078125,
      "learning_rate": 7.669004379373284e-06,
      "loss": 86.6673,
      "step": 48100
    },
    {
      "epoch": 0.39798155271539065,
      "grad_norm": 1260.71875,
      "learning_rate": 7.667811652414637e-06,
      "loss": 87.2874,
      "step": 48110
    },
    {
      "epoch": 0.3980642759647599,
      "grad_norm": 986.4711303710938,
      "learning_rate": 7.666618713190459e-06,
      "loss": 98.885,
      "step": 48120
    },
    {
      "epoch": 0.3981469992141291,
      "grad_norm": 1163.977783203125,
      "learning_rate": 7.665425561795669e-06,
      "loss": 86.0785,
      "step": 48130
    },
    {
      "epoch": 0.39822972246349836,
      "grad_norm": 851.3588256835938,
      "learning_rate": 7.664232198325198e-06,
      "loss": 88.372,
      "step": 48140
    },
    {
      "epoch": 0.3983124457128676,
      "grad_norm": 786.80419921875,
      "learning_rate": 7.663038622873999e-06,
      "loss": 83.9941,
      "step": 48150
    },
    {
      "epoch": 0.3983951689622368,
      "grad_norm": 1034.7119140625,
      "learning_rate": 7.66184483553704e-06,
      "loss": 91.0622,
      "step": 48160
    },
    {
      "epoch": 0.39847789221160607,
      "grad_norm": 1714.0181884765625,
      "learning_rate": 7.660650836409302e-06,
      "loss": 88.6052,
      "step": 48170
    },
    {
      "epoch": 0.3985606154609753,
      "grad_norm": 1282.2288818359375,
      "learning_rate": 7.65945662558579e-06,
      "loss": 102.4522,
      "step": 48180
    },
    {
      "epoch": 0.3986433387103445,
      "grad_norm": 1070.42626953125,
      "learning_rate": 7.658262203161517e-06,
      "loss": 120.5579,
      "step": 48190
    },
    {
      "epoch": 0.3987260619597138,
      "grad_norm": 1412.6171875,
      "learning_rate": 7.65706756923152e-06,
      "loss": 88.6738,
      "step": 48200
    },
    {
      "epoch": 0.398808785209083,
      "grad_norm": 745.8485107421875,
      "learning_rate": 7.655872723890854e-06,
      "loss": 116.865,
      "step": 48210
    },
    {
      "epoch": 0.3988915084584522,
      "grad_norm": 938.5888671875,
      "learning_rate": 7.654677667234582e-06,
      "loss": 112.7861,
      "step": 48220
    },
    {
      "epoch": 0.3989742317078215,
      "grad_norm": 692.1904296875,
      "learning_rate": 7.65348239935779e-06,
      "loss": 93.3395,
      "step": 48230
    },
    {
      "epoch": 0.39905695495719073,
      "grad_norm": 892.9244384765625,
      "learning_rate": 7.652286920355583e-06,
      "loss": 123.322,
      "step": 48240
    },
    {
      "epoch": 0.39913967820655993,
      "grad_norm": 353.6568603515625,
      "learning_rate": 7.651091230323079e-06,
      "loss": 75.8525,
      "step": 48250
    },
    {
      "epoch": 0.3992224014559292,
      "grad_norm": 1054.718017578125,
      "learning_rate": 7.649895329355411e-06,
      "loss": 144.2353,
      "step": 48260
    },
    {
      "epoch": 0.39930512470529844,
      "grad_norm": 854.1619873046875,
      "learning_rate": 7.648699217547733e-06,
      "loss": 89.6152,
      "step": 48270
    },
    {
      "epoch": 0.39938784795466764,
      "grad_norm": 1015.1759033203125,
      "learning_rate": 7.647502894995215e-06,
      "loss": 97.5599,
      "step": 48280
    },
    {
      "epoch": 0.3994705712040369,
      "grad_norm": 691.9119873046875,
      "learning_rate": 7.646306361793042e-06,
      "loss": 95.9693,
      "step": 48290
    },
    {
      "epoch": 0.39955329445340615,
      "grad_norm": 545.0184936523438,
      "learning_rate": 7.645109618036416e-06,
      "loss": 89.2513,
      "step": 48300
    },
    {
      "epoch": 0.39963601770277535,
      "grad_norm": 1180.732666015625,
      "learning_rate": 7.643912663820559e-06,
      "loss": 107.0819,
      "step": 48310
    },
    {
      "epoch": 0.3997187409521446,
      "grad_norm": 634.8629760742188,
      "learning_rate": 7.642715499240702e-06,
      "loss": 93.8869,
      "step": 48320
    },
    {
      "epoch": 0.39980146420151386,
      "grad_norm": 878.6117553710938,
      "learning_rate": 7.641518124392105e-06,
      "loss": 117.2257,
      "step": 48330
    },
    {
      "epoch": 0.39988418745088306,
      "grad_norm": 741.9995727539062,
      "learning_rate": 7.640320539370032e-06,
      "loss": 92.1223,
      "step": 48340
    },
    {
      "epoch": 0.3999669107002523,
      "grad_norm": 693.643310546875,
      "learning_rate": 7.63912274426977e-06,
      "loss": 102.7763,
      "step": 48350
    },
    {
      "epoch": 0.40004963394962156,
      "grad_norm": 747.3237915039062,
      "learning_rate": 7.637924739186624e-06,
      "loss": 83.4088,
      "step": 48360
    },
    {
      "epoch": 0.40013235719899076,
      "grad_norm": 1104.4271240234375,
      "learning_rate": 7.636726524215913e-06,
      "loss": 96.2126,
      "step": 48370
    },
    {
      "epoch": 0.40021508044836,
      "grad_norm": 1066.0506591796875,
      "learning_rate": 7.635528099452974e-06,
      "loss": 137.2111,
      "step": 48380
    },
    {
      "epoch": 0.40029780369772927,
      "grad_norm": 764.3858032226562,
      "learning_rate": 7.634329464993158e-06,
      "loss": 99.6301,
      "step": 48390
    },
    {
      "epoch": 0.40038052694709847,
      "grad_norm": 619.43017578125,
      "learning_rate": 7.633130620931837e-06,
      "loss": 95.3814,
      "step": 48400
    },
    {
      "epoch": 0.4004632501964677,
      "grad_norm": 576.7614135742188,
      "learning_rate": 7.631931567364398e-06,
      "loss": 115.1573,
      "step": 48410
    },
    {
      "epoch": 0.400545973445837,
      "grad_norm": 1004.3753662109375,
      "learning_rate": 7.630732304386244e-06,
      "loss": 90.6397,
      "step": 48420
    },
    {
      "epoch": 0.4006286966952062,
      "grad_norm": 1928.7547607421875,
      "learning_rate": 7.629532832092792e-06,
      "loss": 98.6528,
      "step": 48430
    },
    {
      "epoch": 0.40071141994457543,
      "grad_norm": 1009.5800170898438,
      "learning_rate": 7.62833315057948e-06,
      "loss": 105.3359,
      "step": 48440
    },
    {
      "epoch": 0.4007941431939447,
      "grad_norm": 573.4387817382812,
      "learning_rate": 7.627133259941762e-06,
      "loss": 95.6264,
      "step": 48450
    },
    {
      "epoch": 0.4008768664433139,
      "grad_norm": 1055.6937255859375,
      "learning_rate": 7.625933160275109e-06,
      "loss": 98.1826,
      "step": 48460
    },
    {
      "epoch": 0.40095958969268314,
      "grad_norm": 636.2737426757812,
      "learning_rate": 7.6247328516750055e-06,
      "loss": 106.1883,
      "step": 48470
    },
    {
      "epoch": 0.40104231294205234,
      "grad_norm": 624.5562744140625,
      "learning_rate": 7.623532334236954e-06,
      "loss": 115.2045,
      "step": 48480
    },
    {
      "epoch": 0.4011250361914216,
      "grad_norm": 1036.740234375,
      "learning_rate": 7.622331608056474e-06,
      "loss": 100.7731,
      "step": 48490
    },
    {
      "epoch": 0.40120775944079085,
      "grad_norm": 1890.8753662109375,
      "learning_rate": 7.621130673229105e-06,
      "loss": 98.8333,
      "step": 48500
    },
    {
      "epoch": 0.40129048269016004,
      "grad_norm": 967.46044921875,
      "learning_rate": 7.619929529850397e-06,
      "loss": 62.4726,
      "step": 48510
    },
    {
      "epoch": 0.4013732059395293,
      "grad_norm": 861.5704956054688,
      "learning_rate": 7.618728178015919e-06,
      "loss": 94.6595,
      "step": 48520
    },
    {
      "epoch": 0.40145592918889855,
      "grad_norm": 1017.3840942382812,
      "learning_rate": 7.617526617821259e-06,
      "loss": 107.7875,
      "step": 48530
    },
    {
      "epoch": 0.40153865243826775,
      "grad_norm": 2595.431640625,
      "learning_rate": 7.616324849362019e-06,
      "loss": 104.2326,
      "step": 48540
    },
    {
      "epoch": 0.401621375687637,
      "grad_norm": 767.1851196289062,
      "learning_rate": 7.615122872733819e-06,
      "loss": 126.0866,
      "step": 48550
    },
    {
      "epoch": 0.40170409893700626,
      "grad_norm": 901.1397705078125,
      "learning_rate": 7.613920688032293e-06,
      "loss": 120.4315,
      "step": 48560
    },
    {
      "epoch": 0.40178682218637546,
      "grad_norm": 1408.295166015625,
      "learning_rate": 7.612718295353094e-06,
      "loss": 104.612,
      "step": 48570
    },
    {
      "epoch": 0.4018695454357447,
      "grad_norm": 746.8675537109375,
      "learning_rate": 7.61151569479189e-06,
      "loss": 84.4078,
      "step": 48580
    },
    {
      "epoch": 0.40195226868511397,
      "grad_norm": 925.1431884765625,
      "learning_rate": 7.610312886444369e-06,
      "loss": 89.8368,
      "step": 48590
    },
    {
      "epoch": 0.40203499193448317,
      "grad_norm": 757.10693359375,
      "learning_rate": 7.60910987040623e-06,
      "loss": 92.3049,
      "step": 48600
    },
    {
      "epoch": 0.4021177151838524,
      "grad_norm": 777.216552734375,
      "learning_rate": 7.607906646773195e-06,
      "loss": 105.3716,
      "step": 48610
    },
    {
      "epoch": 0.4022004384332217,
      "grad_norm": 751.5682983398438,
      "learning_rate": 7.606703215640995e-06,
      "loss": 92.2909,
      "step": 48620
    },
    {
      "epoch": 0.4022831616825909,
      "grad_norm": 1239.3206787109375,
      "learning_rate": 7.605499577105382e-06,
      "loss": 106.9025,
      "step": 48630
    },
    {
      "epoch": 0.4023658849319601,
      "grad_norm": 750.9842529296875,
      "learning_rate": 7.604295731262128e-06,
      "loss": 115.5005,
      "step": 48640
    },
    {
      "epoch": 0.4024486081813294,
      "grad_norm": 1378.44677734375,
      "learning_rate": 7.603091678207013e-06,
      "loss": 95.4831,
      "step": 48650
    },
    {
      "epoch": 0.4025313314306986,
      "grad_norm": 883.4937133789062,
      "learning_rate": 7.60188741803584e-06,
      "loss": 94.2378,
      "step": 48660
    },
    {
      "epoch": 0.40261405468006783,
      "grad_norm": 927.4861450195312,
      "learning_rate": 7.600682950844428e-06,
      "loss": 97.9098,
      "step": 48670
    },
    {
      "epoch": 0.4026967779294371,
      "grad_norm": 999.6773071289062,
      "learning_rate": 7.599478276728607e-06,
      "loss": 88.1278,
      "step": 48680
    },
    {
      "epoch": 0.4027795011788063,
      "grad_norm": 981.5151977539062,
      "learning_rate": 7.5982733957842304e-06,
      "loss": 96.1045,
      "step": 48690
    },
    {
      "epoch": 0.40286222442817554,
      "grad_norm": 779.521728515625,
      "learning_rate": 7.597068308107165e-06,
      "loss": 96.5194,
      "step": 48700
    },
    {
      "epoch": 0.4029449476775448,
      "grad_norm": 723.9231567382812,
      "learning_rate": 7.595863013793292e-06,
      "loss": 105.2723,
      "step": 48710
    },
    {
      "epoch": 0.403027670926914,
      "grad_norm": 688.5134887695312,
      "learning_rate": 7.594657512938513e-06,
      "loss": 73.8923,
      "step": 48720
    },
    {
      "epoch": 0.40311039417628325,
      "grad_norm": 1107.6732177734375,
      "learning_rate": 7.593451805638743e-06,
      "loss": 84.0132,
      "step": 48730
    },
    {
      "epoch": 0.4031931174256525,
      "grad_norm": 971.2855834960938,
      "learning_rate": 7.592245891989914e-06,
      "loss": 126.4593,
      "step": 48740
    },
    {
      "epoch": 0.4032758406750217,
      "grad_norm": 906.73388671875,
      "learning_rate": 7.5910397720879785e-06,
      "loss": 112.5872,
      "step": 48750
    },
    {
      "epoch": 0.40335856392439096,
      "grad_norm": 976.5882568359375,
      "learning_rate": 7.589833446028898e-06,
      "loss": 113.1635,
      "step": 48760
    },
    {
      "epoch": 0.4034412871737602,
      "grad_norm": 517.5700073242188,
      "learning_rate": 7.5886269139086565e-06,
      "loss": 95.9183,
      "step": 48770
    },
    {
      "epoch": 0.4035240104231294,
      "grad_norm": 1049.484375,
      "learning_rate": 7.587420175823252e-06,
      "loss": 114.5341,
      "step": 48780
    },
    {
      "epoch": 0.40360673367249866,
      "grad_norm": 1442.658447265625,
      "learning_rate": 7.586213231868699e-06,
      "loss": 89.2601,
      "step": 48790
    },
    {
      "epoch": 0.4036894569218679,
      "grad_norm": 923.6640625,
      "learning_rate": 7.585006082141028e-06,
      "loss": 100.5833,
      "step": 48800
    },
    {
      "epoch": 0.4037721801712371,
      "grad_norm": 882.6138916015625,
      "learning_rate": 7.583798726736286e-06,
      "loss": 111.9895,
      "step": 48810
    },
    {
      "epoch": 0.40385490342060637,
      "grad_norm": 841.87548828125,
      "learning_rate": 7.5825911657505365e-06,
      "loss": 110.3644,
      "step": 48820
    },
    {
      "epoch": 0.40393762666997557,
      "grad_norm": 686.5569458007812,
      "learning_rate": 7.581383399279863e-06,
      "loss": 113.7908,
      "step": 48830
    },
    {
      "epoch": 0.4040203499193448,
      "grad_norm": 1229.239501953125,
      "learning_rate": 7.580175427420358e-06,
      "loss": 94.2136,
      "step": 48840
    },
    {
      "epoch": 0.4041030731687141,
      "grad_norm": 1013.9456176757812,
      "learning_rate": 7.578967250268137e-06,
      "loss": 112.4359,
      "step": 48850
    },
    {
      "epoch": 0.4041857964180833,
      "grad_norm": 843.6847534179688,
      "learning_rate": 7.577758867919325e-06,
      "loss": 119.3482,
      "step": 48860
    },
    {
      "epoch": 0.40426851966745253,
      "grad_norm": 660.8492431640625,
      "learning_rate": 7.576550280470072e-06,
      "loss": 105.5132,
      "step": 48870
    },
    {
      "epoch": 0.4043512429168218,
      "grad_norm": 1110.703125,
      "learning_rate": 7.5753414880165365e-06,
      "loss": 117.6248,
      "step": 48880
    },
    {
      "epoch": 0.404433966166191,
      "grad_norm": 971.7891235351562,
      "learning_rate": 7.5741324906548996e-06,
      "loss": 84.8425,
      "step": 48890
    },
    {
      "epoch": 0.40451668941556024,
      "grad_norm": 674.7775268554688,
      "learning_rate": 7.572923288481355e-06,
      "loss": 93.9152,
      "step": 48900
    },
    {
      "epoch": 0.4045994126649295,
      "grad_norm": 1152.0150146484375,
      "learning_rate": 7.571713881592109e-06,
      "loss": 97.321,
      "step": 48910
    },
    {
      "epoch": 0.4046821359142987,
      "grad_norm": 813.1100463867188,
      "learning_rate": 7.570504270083394e-06,
      "loss": 107.887,
      "step": 48920
    },
    {
      "epoch": 0.40476485916366794,
      "grad_norm": 860.7267456054688,
      "learning_rate": 7.569294454051452e-06,
      "loss": 93.7367,
      "step": 48930
    },
    {
      "epoch": 0.4048475824130372,
      "grad_norm": 641.8156127929688,
      "learning_rate": 7.568084433592542e-06,
      "loss": 103.0248,
      "step": 48940
    },
    {
      "epoch": 0.4049303056624064,
      "grad_norm": 1134.04736328125,
      "learning_rate": 7.566874208802939e-06,
      "loss": 101.0284,
      "step": 48950
    },
    {
      "epoch": 0.40501302891177565,
      "grad_norm": 596.593017578125,
      "learning_rate": 7.5656637797789335e-06,
      "loss": 80.7004,
      "step": 48960
    },
    {
      "epoch": 0.4050957521611449,
      "grad_norm": 791.4570922851562,
      "learning_rate": 7.564453146616837e-06,
      "loss": 109.8588,
      "step": 48970
    },
    {
      "epoch": 0.4051784754105141,
      "grad_norm": 681.3819580078125,
      "learning_rate": 7.563242309412975e-06,
      "loss": 105.6946,
      "step": 48980
    },
    {
      "epoch": 0.40526119865988336,
      "grad_norm": 881.9998168945312,
      "learning_rate": 7.562031268263686e-06,
      "loss": 104.3682,
      "step": 48990
    },
    {
      "epoch": 0.4053439219092526,
      "grad_norm": 985.5377807617188,
      "learning_rate": 7.5608200232653254e-06,
      "loss": 146.3788,
      "step": 49000
    },
    {
      "epoch": 0.4054266451586218,
      "grad_norm": 547.89697265625,
      "learning_rate": 7.5596085745142654e-06,
      "loss": 126.0642,
      "step": 49010
    },
    {
      "epoch": 0.40550936840799107,
      "grad_norm": 1244.3304443359375,
      "learning_rate": 7.558396922106903e-06,
      "loss": 105.9497,
      "step": 49020
    },
    {
      "epoch": 0.4055920916573603,
      "grad_norm": 1159.8284912109375,
      "learning_rate": 7.557185066139638e-06,
      "loss": 123.0697,
      "step": 49030
    },
    {
      "epoch": 0.4056748149067295,
      "grad_norm": 1442.5506591796875,
      "learning_rate": 7.555973006708892e-06,
      "loss": 124.4241,
      "step": 49040
    },
    {
      "epoch": 0.4057575381560988,
      "grad_norm": 1344.17236328125,
      "learning_rate": 7.554760743911104e-06,
      "loss": 98.3027,
      "step": 49050
    },
    {
      "epoch": 0.40584026140546803,
      "grad_norm": 733.8927001953125,
      "learning_rate": 7.553548277842729e-06,
      "loss": 90.8391,
      "step": 49060
    },
    {
      "epoch": 0.4059229846548372,
      "grad_norm": 1737.365966796875,
      "learning_rate": 7.5523356086002364e-06,
      "loss": 128.2919,
      "step": 49070
    },
    {
      "epoch": 0.4060057079042065,
      "grad_norm": 1121.77783203125,
      "learning_rate": 7.551122736280113e-06,
      "loss": 114.6872,
      "step": 49080
    },
    {
      "epoch": 0.40608843115357574,
      "grad_norm": 830.7010498046875,
      "learning_rate": 7.549909660978863e-06,
      "loss": 103.0037,
      "step": 49090
    },
    {
      "epoch": 0.40617115440294493,
      "grad_norm": 950.9368896484375,
      "learning_rate": 7.548696382793002e-06,
      "loss": 76.3594,
      "step": 49100
    },
    {
      "epoch": 0.4062538776523142,
      "grad_norm": 721.6538696289062,
      "learning_rate": 7.547482901819066e-06,
      "loss": 101.0682,
      "step": 49110
    },
    {
      "epoch": 0.40633660090168344,
      "grad_norm": 1179.335693359375,
      "learning_rate": 7.5462692181536094e-06,
      "loss": 113.5078,
      "step": 49120
    },
    {
      "epoch": 0.40641932415105264,
      "grad_norm": 1521.753173828125,
      "learning_rate": 7.545055331893195e-06,
      "loss": 127.0566,
      "step": 49130
    },
    {
      "epoch": 0.4065020474004219,
      "grad_norm": 718.9688720703125,
      "learning_rate": 7.543841243134409e-06,
      "loss": 77.3431,
      "step": 49140
    },
    {
      "epoch": 0.40658477064979115,
      "grad_norm": 1070.0787353515625,
      "learning_rate": 7.5426269519738495e-06,
      "loss": 108.8352,
      "step": 49150
    },
    {
      "epoch": 0.40666749389916035,
      "grad_norm": 514.2880859375,
      "learning_rate": 7.541412458508133e-06,
      "loss": 104.5422,
      "step": 49160
    },
    {
      "epoch": 0.4067502171485296,
      "grad_norm": 774.6307983398438,
      "learning_rate": 7.54019776283389e-06,
      "loss": 100.5812,
      "step": 49170
    },
    {
      "epoch": 0.40683294039789886,
      "grad_norm": 600.9510498046875,
      "learning_rate": 7.53898286504777e-06,
      "loss": 75.3506,
      "step": 49180
    },
    {
      "epoch": 0.40691566364726806,
      "grad_norm": 843.1857299804688,
      "learning_rate": 7.537767765246436e-06,
      "loss": 110.9489,
      "step": 49190
    },
    {
      "epoch": 0.4069983868966373,
      "grad_norm": 957.2750854492188,
      "learning_rate": 7.536552463526565e-06,
      "loss": 86.3647,
      "step": 49200
    },
    {
      "epoch": 0.4070811101460065,
      "grad_norm": 1061.4150390625,
      "learning_rate": 7.535336959984858e-06,
      "loss": 127.3531,
      "step": 49210
    },
    {
      "epoch": 0.40716383339537576,
      "grad_norm": 2062.2001953125,
      "learning_rate": 7.5341212547180246e-06,
      "loss": 111.2139,
      "step": 49220
    },
    {
      "epoch": 0.407246556644745,
      "grad_norm": 768.780517578125,
      "learning_rate": 7.532905347822792e-06,
      "loss": 116.6353,
      "step": 49230
    },
    {
      "epoch": 0.4073292798941142,
      "grad_norm": 780.7527465820312,
      "learning_rate": 7.5316892393959064e-06,
      "loss": 91.1807,
      "step": 49240
    },
    {
      "epoch": 0.40741200314348347,
      "grad_norm": 487.50665283203125,
      "learning_rate": 7.530472929534126e-06,
      "loss": 110.9754,
      "step": 49250
    },
    {
      "epoch": 0.4074947263928527,
      "grad_norm": 578.7011108398438,
      "learning_rate": 7.529256418334228e-06,
      "loss": 131.2194,
      "step": 49260
    },
    {
      "epoch": 0.4075774496422219,
      "grad_norm": 818.4442749023438,
      "learning_rate": 7.528039705893006e-06,
      "loss": 98.4813,
      "step": 49270
    },
    {
      "epoch": 0.4076601728915912,
      "grad_norm": 555.372314453125,
      "learning_rate": 7.5268227923072665e-06,
      "loss": 100.5567,
      "step": 49280
    },
    {
      "epoch": 0.40774289614096043,
      "grad_norm": 1478.378173828125,
      "learning_rate": 7.525605677673831e-06,
      "loss": 94.0006,
      "step": 49290
    },
    {
      "epoch": 0.40782561939032963,
      "grad_norm": 1054.407958984375,
      "learning_rate": 7.524388362089545e-06,
      "loss": 110.3638,
      "step": 49300
    },
    {
      "epoch": 0.4079083426396989,
      "grad_norm": 936.5081176757812,
      "learning_rate": 7.523170845651263e-06,
      "loss": 108.1984,
      "step": 49310
    },
    {
      "epoch": 0.40799106588906814,
      "grad_norm": 714.961181640625,
      "learning_rate": 7.521953128455856e-06,
      "loss": 109.1958,
      "step": 49320
    },
    {
      "epoch": 0.40807378913843734,
      "grad_norm": 850.507080078125,
      "learning_rate": 7.520735210600213e-06,
      "loss": 117.9377,
      "step": 49330
    },
    {
      "epoch": 0.4081565123878066,
      "grad_norm": 815.2796020507812,
      "learning_rate": 7.519517092181237e-06,
      "loss": 128.2985,
      "step": 49340
    },
    {
      "epoch": 0.40823923563717585,
      "grad_norm": 1209.8961181640625,
      "learning_rate": 7.518298773295849e-06,
      "loss": 108.5808,
      "step": 49350
    },
    {
      "epoch": 0.40832195888654504,
      "grad_norm": 1075.5911865234375,
      "learning_rate": 7.517080254040985e-06,
      "loss": 87.4483,
      "step": 49360
    },
    {
      "epoch": 0.4084046821359143,
      "grad_norm": 952.4523315429688,
      "learning_rate": 7.5158615345136e-06,
      "loss": 85.5586,
      "step": 49370
    },
    {
      "epoch": 0.40848740538528355,
      "grad_norm": 688.7626342773438,
      "learning_rate": 7.514642614810655e-06,
      "loss": 113.6158,
      "step": 49380
    },
    {
      "epoch": 0.40857012863465275,
      "grad_norm": 754.0592651367188,
      "learning_rate": 7.51342349502914e-06,
      "loss": 95.794,
      "step": 49390
    },
    {
      "epoch": 0.408652851884022,
      "grad_norm": 1221.88818359375,
      "learning_rate": 7.512204175266052e-06,
      "loss": 99.691,
      "step": 49400
    },
    {
      "epoch": 0.40873557513339126,
      "grad_norm": 781.3279418945312,
      "learning_rate": 7.510984655618407e-06,
      "loss": 105.6204,
      "step": 49410
    },
    {
      "epoch": 0.40881829838276046,
      "grad_norm": 593.7680053710938,
      "learning_rate": 7.509764936183237e-06,
      "loss": 82.4356,
      "step": 49420
    },
    {
      "epoch": 0.4089010216321297,
      "grad_norm": 2055.203369140625,
      "learning_rate": 7.5085450170575876e-06,
      "loss": 120.1463,
      "step": 49430
    },
    {
      "epoch": 0.40898374488149897,
      "grad_norm": 1296.7412109375,
      "learning_rate": 7.5073248983385265e-06,
      "loss": 130.0576,
      "step": 49440
    },
    {
      "epoch": 0.40906646813086817,
      "grad_norm": 754.3132934570312,
      "learning_rate": 7.50610458012313e-06,
      "loss": 88.8123,
      "step": 49450
    },
    {
      "epoch": 0.4091491913802374,
      "grad_norm": 827.0701904296875,
      "learning_rate": 7.504884062508493e-06,
      "loss": 85.6176,
      "step": 49460
    },
    {
      "epoch": 0.4092319146296067,
      "grad_norm": 533.4575805664062,
      "learning_rate": 7.503663345591726e-06,
      "loss": 99.3742,
      "step": 49470
    },
    {
      "epoch": 0.4093146378789759,
      "grad_norm": 914.8536376953125,
      "learning_rate": 7.502442429469956e-06,
      "loss": 123.0487,
      "step": 49480
    },
    {
      "epoch": 0.4093973611283451,
      "grad_norm": 1090.47998046875,
      "learning_rate": 7.501221314240329e-06,
      "loss": 110.9421,
      "step": 49490
    },
    {
      "epoch": 0.4094800843777144,
      "grad_norm": 969.5048828125,
      "learning_rate": 7.500000000000001e-06,
      "loss": 132.2819,
      "step": 49500
    },
    {
      "epoch": 0.4095628076270836,
      "grad_norm": 1003.5017700195312,
      "learning_rate": 7.4987784868461455e-06,
      "loss": 125.4059,
      "step": 49510
    },
    {
      "epoch": 0.40964553087645283,
      "grad_norm": 1233.950927734375,
      "learning_rate": 7.497556774875953e-06,
      "loss": 100.2119,
      "step": 49520
    },
    {
      "epoch": 0.4097282541258221,
      "grad_norm": 638.7094116210938,
      "learning_rate": 7.496334864186632e-06,
      "loss": 133.3458,
      "step": 49530
    },
    {
      "epoch": 0.4098109773751913,
      "grad_norm": 448.5832214355469,
      "learning_rate": 7.4951127548754025e-06,
      "loss": 95.0754,
      "step": 49540
    },
    {
      "epoch": 0.40989370062456054,
      "grad_norm": 1257.3818359375,
      "learning_rate": 7.4938904470395e-06,
      "loss": 91.5809,
      "step": 49550
    },
    {
      "epoch": 0.40997642387392974,
      "grad_norm": 1006.02294921875,
      "learning_rate": 7.492667940776182e-06,
      "loss": 102.9834,
      "step": 49560
    },
    {
      "epoch": 0.410059147123299,
      "grad_norm": 1712.72607421875,
      "learning_rate": 7.491445236182715e-06,
      "loss": 103.0404,
      "step": 49570
    },
    {
      "epoch": 0.41014187037266825,
      "grad_norm": 799.1417236328125,
      "learning_rate": 7.490222333356384e-06,
      "loss": 106.2483,
      "step": 49580
    },
    {
      "epoch": 0.41022459362203745,
      "grad_norm": 1142.0809326171875,
      "learning_rate": 7.488999232394492e-06,
      "loss": 151.4166,
      "step": 49590
    },
    {
      "epoch": 0.4103073168714067,
      "grad_norm": 1388.60693359375,
      "learning_rate": 7.487775933394353e-06,
      "loss": 103.9434,
      "step": 49600
    },
    {
      "epoch": 0.41039004012077596,
      "grad_norm": 835.6099853515625,
      "learning_rate": 7.4865524364533e-06,
      "loss": 104.3207,
      "step": 49610
    },
    {
      "epoch": 0.41047276337014516,
      "grad_norm": 653.7705688476562,
      "learning_rate": 7.485328741668683e-06,
      "loss": 83.743,
      "step": 49620
    },
    {
      "epoch": 0.4105554866195144,
      "grad_norm": 870.46484375,
      "learning_rate": 7.484104849137862e-06,
      "loss": 108.1679,
      "step": 49630
    },
    {
      "epoch": 0.41063820986888366,
      "grad_norm": 839.9007568359375,
      "learning_rate": 7.482880758958219e-06,
      "loss": 83.2425,
      "step": 49640
    },
    {
      "epoch": 0.41072093311825286,
      "grad_norm": 931.4784545898438,
      "learning_rate": 7.48165647122715e-06,
      "loss": 114.039,
      "step": 49650
    },
    {
      "epoch": 0.4108036563676221,
      "grad_norm": 776.007080078125,
      "learning_rate": 7.480431986042065e-06,
      "loss": 138.1686,
      "step": 49660
    },
    {
      "epoch": 0.41088637961699137,
      "grad_norm": 890.8333129882812,
      "learning_rate": 7.47920730350039e-06,
      "loss": 129.4647,
      "step": 49670
    },
    {
      "epoch": 0.41096910286636057,
      "grad_norm": 1296.5220947265625,
      "learning_rate": 7.477982423699568e-06,
      "loss": 120.0857,
      "step": 49680
    },
    {
      "epoch": 0.4110518261157298,
      "grad_norm": 1131.2828369140625,
      "learning_rate": 7.476757346737057e-06,
      "loss": 112.3677,
      "step": 49690
    },
    {
      "epoch": 0.4111345493650991,
      "grad_norm": 1283.157470703125,
      "learning_rate": 7.47553207271033e-06,
      "loss": 114.4162,
      "step": 49700
    },
    {
      "epoch": 0.4112172726144683,
      "grad_norm": 309.3968200683594,
      "learning_rate": 7.474306601716877e-06,
      "loss": 89.349,
      "step": 49710
    },
    {
      "epoch": 0.41129999586383753,
      "grad_norm": 961.7044677734375,
      "learning_rate": 7.473080933854205e-06,
      "loss": 81.6981,
      "step": 49720
    },
    {
      "epoch": 0.4113827191132068,
      "grad_norm": 427.75469970703125,
      "learning_rate": 7.471855069219831e-06,
      "loss": 111.079,
      "step": 49730
    },
    {
      "epoch": 0.411465442362576,
      "grad_norm": 973.0006103515625,
      "learning_rate": 7.470629007911294e-06,
      "loss": 99.5809,
      "step": 49740
    },
    {
      "epoch": 0.41154816561194524,
      "grad_norm": 1471.537109375,
      "learning_rate": 7.469402750026147e-06,
      "loss": 130.2198,
      "step": 49750
    },
    {
      "epoch": 0.4116308888613145,
      "grad_norm": 765.6078491210938,
      "learning_rate": 7.468176295661955e-06,
      "loss": 124.7598,
      "step": 49760
    },
    {
      "epoch": 0.4117136121106837,
      "grad_norm": 918.6041259765625,
      "learning_rate": 7.466949644916301e-06,
      "loss": 103.558,
      "step": 49770
    },
    {
      "epoch": 0.41179633536005295,
      "grad_norm": 933.6361694335938,
      "learning_rate": 7.465722797886788e-06,
      "loss": 94.5863,
      "step": 49780
    },
    {
      "epoch": 0.4118790586094222,
      "grad_norm": 623.9163818359375,
      "learning_rate": 7.464495754671027e-06,
      "loss": 86.9486,
      "step": 49790
    },
    {
      "epoch": 0.4119617818587914,
      "grad_norm": 592.259033203125,
      "learning_rate": 7.4632685153666505e-06,
      "loss": 111.5722,
      "step": 49800
    },
    {
      "epoch": 0.41204450510816065,
      "grad_norm": 430.714599609375,
      "learning_rate": 7.462041080071301e-06,
      "loss": 81.418,
      "step": 49810
    },
    {
      "epoch": 0.4121272283575299,
      "grad_norm": 1098.07373046875,
      "learning_rate": 7.460813448882643e-06,
      "loss": 145.5894,
      "step": 49820
    },
    {
      "epoch": 0.4122099516068991,
      "grad_norm": 630.3162841796875,
      "learning_rate": 7.459585621898353e-06,
      "loss": 88.393,
      "step": 49830
    },
    {
      "epoch": 0.41229267485626836,
      "grad_norm": 783.9848022460938,
      "learning_rate": 7.4583575992161235e-06,
      "loss": 73.975,
      "step": 49840
    },
    {
      "epoch": 0.4123753981056376,
      "grad_norm": 645.27392578125,
      "learning_rate": 7.457129380933662e-06,
      "loss": 108.0057,
      "step": 49850
    },
    {
      "epoch": 0.4124581213550068,
      "grad_norm": 522.0549926757812,
      "learning_rate": 7.4559009671486906e-06,
      "loss": 65.7118,
      "step": 49860
    },
    {
      "epoch": 0.41254084460437607,
      "grad_norm": 706.9224853515625,
      "learning_rate": 7.454672357958951e-06,
      "loss": 127.6793,
      "step": 49870
    },
    {
      "epoch": 0.4126235678537453,
      "grad_norm": 1543.4708251953125,
      "learning_rate": 7.453443553462198e-06,
      "loss": 138.2186,
      "step": 49880
    },
    {
      "epoch": 0.4127062911031145,
      "grad_norm": 1025.1986083984375,
      "learning_rate": 7.4522145537562015e-06,
      "loss": 117.5677,
      "step": 49890
    },
    {
      "epoch": 0.4127890143524838,
      "grad_norm": 611.4989013671875,
      "learning_rate": 7.450985358938747e-06,
      "loss": 106.0235,
      "step": 49900
    },
    {
      "epoch": 0.41287173760185303,
      "grad_norm": 1154.43701171875,
      "learning_rate": 7.449755969107635e-06,
      "loss": 113.3347,
      "step": 49910
    },
    {
      "epoch": 0.4129544608512222,
      "grad_norm": 1076.0106201171875,
      "learning_rate": 7.4485263843606835e-06,
      "loss": 104.9749,
      "step": 49920
    },
    {
      "epoch": 0.4130371841005915,
      "grad_norm": 1215.836669921875,
      "learning_rate": 7.447296604795726e-06,
      "loss": 119.1512,
      "step": 49930
    },
    {
      "epoch": 0.4131199073499607,
      "grad_norm": 1207.3167724609375,
      "learning_rate": 7.4460666305106084e-06,
      "loss": 101.0483,
      "step": 49940
    },
    {
      "epoch": 0.41320263059932993,
      "grad_norm": 1147.7183837890625,
      "learning_rate": 7.444836461603195e-06,
      "loss": 101.2153,
      "step": 49950
    },
    {
      "epoch": 0.4132853538486992,
      "grad_norm": 784.3466796875,
      "learning_rate": 7.443606098171363e-06,
      "loss": 86.7002,
      "step": 49960
    },
    {
      "epoch": 0.4133680770980684,
      "grad_norm": 452.4320068359375,
      "learning_rate": 7.442375540313012e-06,
      "loss": 94.4948,
      "step": 49970
    },
    {
      "epoch": 0.41345080034743764,
      "grad_norm": 911.47021484375,
      "learning_rate": 7.441144788126045e-06,
      "loss": 110.6758,
      "step": 49980
    },
    {
      "epoch": 0.4135335235968069,
      "grad_norm": 1147.6248779296875,
      "learning_rate": 7.4399138417083925e-06,
      "loss": 95.634,
      "step": 49990
    },
    {
      "epoch": 0.4136162468461761,
      "grad_norm": 1253.6624755859375,
      "learning_rate": 7.438682701157993e-06,
      "loss": 141.2699,
      "step": 50000
    },
    {
      "epoch": 0.41369897009554535,
      "grad_norm": 1440.07080078125,
      "learning_rate": 7.437451366572803e-06,
      "loss": 109.6948,
      "step": 50010
    },
    {
      "epoch": 0.4137816933449146,
      "grad_norm": 833.8212280273438,
      "learning_rate": 7.436219838050793e-06,
      "loss": 106.7741,
      "step": 50020
    },
    {
      "epoch": 0.4138644165942838,
      "grad_norm": 999.9931640625,
      "learning_rate": 7.4349881156899525e-06,
      "loss": 110.1778,
      "step": 50030
    },
    {
      "epoch": 0.41394713984365306,
      "grad_norm": 1517.33251953125,
      "learning_rate": 7.433756199588282e-06,
      "loss": 106.7463,
      "step": 50040
    },
    {
      "epoch": 0.4140298630930223,
      "grad_norm": 749.0157470703125,
      "learning_rate": 7.4325240898438e-06,
      "loss": 86.2277,
      "step": 50050
    },
    {
      "epoch": 0.4141125863423915,
      "grad_norm": 1004.859619140625,
      "learning_rate": 7.4312917865545406e-06,
      "loss": 101.3041,
      "step": 50060
    },
    {
      "epoch": 0.41419530959176076,
      "grad_norm": 2136.435546875,
      "learning_rate": 7.430059289818552e-06,
      "loss": 109.2253,
      "step": 50070
    },
    {
      "epoch": 0.41427803284113,
      "grad_norm": 1305.9910888671875,
      "learning_rate": 7.4288265997338985e-06,
      "loss": 116.3073,
      "step": 50080
    },
    {
      "epoch": 0.4143607560904992,
      "grad_norm": 1031.19921875,
      "learning_rate": 7.427593716398658e-06,
      "loss": 136.2479,
      "step": 50090
    },
    {
      "epoch": 0.41444347933986847,
      "grad_norm": 539.9341430664062,
      "learning_rate": 7.426360639910927e-06,
      "loss": 86.8462,
      "step": 50100
    },
    {
      "epoch": 0.4145262025892377,
      "grad_norm": 527.5219116210938,
      "learning_rate": 7.425127370368815e-06,
      "loss": 104.5311,
      "step": 50110
    },
    {
      "epoch": 0.4146089258386069,
      "grad_norm": 836.485107421875,
      "learning_rate": 7.423893907870449e-06,
      "loss": 125.8744,
      "step": 50120
    },
    {
      "epoch": 0.4146916490879762,
      "grad_norm": 514.3257446289062,
      "learning_rate": 7.422660252513969e-06,
      "loss": 100.2241,
      "step": 50130
    },
    {
      "epoch": 0.41477437233734543,
      "grad_norm": 773.184814453125,
      "learning_rate": 7.421426404397531e-06,
      "loss": 82.5671,
      "step": 50140
    },
    {
      "epoch": 0.41485709558671463,
      "grad_norm": 1118.6546630859375,
      "learning_rate": 7.420192363619305e-06,
      "loss": 116.5798,
      "step": 50150
    },
    {
      "epoch": 0.4149398188360839,
      "grad_norm": 1059.586669921875,
      "learning_rate": 7.418958130277483e-06,
      "loss": 94.7955,
      "step": 50160
    },
    {
      "epoch": 0.41502254208545314,
      "grad_norm": 468.1589660644531,
      "learning_rate": 7.417723704470261e-06,
      "loss": 92.9759,
      "step": 50170
    },
    {
      "epoch": 0.41510526533482234,
      "grad_norm": 2084.325927734375,
      "learning_rate": 7.4164890862958615e-06,
      "loss": 128.8667,
      "step": 50180
    },
    {
      "epoch": 0.4151879885841916,
      "grad_norm": 834.3111572265625,
      "learning_rate": 7.415254275852515e-06,
      "loss": 103.7669,
      "step": 50190
    },
    {
      "epoch": 0.41527071183356085,
      "grad_norm": 869.2041625976562,
      "learning_rate": 7.414019273238471e-06,
      "loss": 88.4897,
      "step": 50200
    },
    {
      "epoch": 0.41535343508293004,
      "grad_norm": 1044.08056640625,
      "learning_rate": 7.4127840785519915e-06,
      "loss": 98.155,
      "step": 50210
    },
    {
      "epoch": 0.4154361583322993,
      "grad_norm": 967.443603515625,
      "learning_rate": 7.411548691891357e-06,
      "loss": 115.068,
      "step": 50220
    },
    {
      "epoch": 0.41551888158166855,
      "grad_norm": 534.2567138671875,
      "learning_rate": 7.41031311335486e-06,
      "loss": 106.5501,
      "step": 50230
    },
    {
      "epoch": 0.41560160483103775,
      "grad_norm": 1074.7396240234375,
      "learning_rate": 7.409077343040809e-06,
      "loss": 153.6957,
      "step": 50240
    },
    {
      "epoch": 0.415684328080407,
      "grad_norm": 657.9391479492188,
      "learning_rate": 7.407841381047533e-06,
      "loss": 99.6328,
      "step": 50250
    },
    {
      "epoch": 0.41576705132977626,
      "grad_norm": 750.8782958984375,
      "learning_rate": 7.406605227473367e-06,
      "loss": 88.5056,
      "step": 50260
    },
    {
      "epoch": 0.41584977457914546,
      "grad_norm": 912.8604125976562,
      "learning_rate": 7.405368882416668e-06,
      "loss": 127.2815,
      "step": 50270
    },
    {
      "epoch": 0.4159324978285147,
      "grad_norm": 722.0275268554688,
      "learning_rate": 7.404132345975806e-06,
      "loss": 89.6634,
      "step": 50280
    },
    {
      "epoch": 0.4160152210778839,
      "grad_norm": 503.6108093261719,
      "learning_rate": 7.4028956182491665e-06,
      "loss": 65.2768,
      "step": 50290
    },
    {
      "epoch": 0.41609794432725317,
      "grad_norm": 1528.93212890625,
      "learning_rate": 7.401658699335151e-06,
      "loss": 110.4871,
      "step": 50300
    },
    {
      "epoch": 0.4161806675766224,
      "grad_norm": 742.7732543945312,
      "learning_rate": 7.400421589332175e-06,
      "loss": 85.2059,
      "step": 50310
    },
    {
      "epoch": 0.4162633908259916,
      "grad_norm": 1395.8526611328125,
      "learning_rate": 7.39918428833867e-06,
      "loss": 94.5206,
      "step": 50320
    },
    {
      "epoch": 0.4163461140753609,
      "grad_norm": 978.8078002929688,
      "learning_rate": 7.397946796453081e-06,
      "loss": 122.651,
      "step": 50330
    },
    {
      "epoch": 0.41642883732473013,
      "grad_norm": 738.7698974609375,
      "learning_rate": 7.39670911377387e-06,
      "loss": 128.4523,
      "step": 50340
    },
    {
      "epoch": 0.4165115605740993,
      "grad_norm": 924.0703735351562,
      "learning_rate": 7.395471240399515e-06,
      "loss": 100.5796,
      "step": 50350
    },
    {
      "epoch": 0.4165942838234686,
      "grad_norm": 1406.5579833984375,
      "learning_rate": 7.394233176428508e-06,
      "loss": 84.9948,
      "step": 50360
    },
    {
      "epoch": 0.41667700707283784,
      "grad_norm": 692.1464233398438,
      "learning_rate": 7.3929949219593545e-06,
      "loss": 108.0571,
      "step": 50370
    },
    {
      "epoch": 0.41675973032220703,
      "grad_norm": 935.2982177734375,
      "learning_rate": 7.391756477090577e-06,
      "loss": 106.4006,
      "step": 50380
    },
    {
      "epoch": 0.4168424535715763,
      "grad_norm": 803.556396484375,
      "learning_rate": 7.3905178419207126e-06,
      "loss": 91.8345,
      "step": 50390
    },
    {
      "epoch": 0.41692517682094554,
      "grad_norm": 170.64480590820312,
      "learning_rate": 7.3892790165483164e-06,
      "loss": 86.5613,
      "step": 50400
    },
    {
      "epoch": 0.41700790007031474,
      "grad_norm": 846.9085693359375,
      "learning_rate": 7.388040001071953e-06,
      "loss": 116.5208,
      "step": 50410
    },
    {
      "epoch": 0.417090623319684,
      "grad_norm": 488.96343994140625,
      "learning_rate": 7.386800795590208e-06,
      "loss": 84.3048,
      "step": 50420
    },
    {
      "epoch": 0.41717334656905325,
      "grad_norm": 2128.975830078125,
      "learning_rate": 7.385561400201675e-06,
      "loss": 101.9137,
      "step": 50430
    },
    {
      "epoch": 0.41725606981842245,
      "grad_norm": 886.4940795898438,
      "learning_rate": 7.384321815004971e-06,
      "loss": 122.7343,
      "step": 50440
    },
    {
      "epoch": 0.4173387930677917,
      "grad_norm": 1239.8963623046875,
      "learning_rate": 7.383082040098723e-06,
      "loss": 118.5375,
      "step": 50450
    },
    {
      "epoch": 0.41742151631716096,
      "grad_norm": 598.2474365234375,
      "learning_rate": 7.381842075581573e-06,
      "loss": 82.4002,
      "step": 50460
    },
    {
      "epoch": 0.41750423956653016,
      "grad_norm": 927.5972900390625,
      "learning_rate": 7.380601921552181e-06,
      "loss": 96.2848,
      "step": 50470
    },
    {
      "epoch": 0.4175869628158994,
      "grad_norm": 1360.900146484375,
      "learning_rate": 7.379361578109218e-06,
      "loss": 99.3107,
      "step": 50480
    },
    {
      "epoch": 0.41766968606526866,
      "grad_norm": 637.6759643554688,
      "learning_rate": 7.378121045351378e-06,
      "loss": 115.3234,
      "step": 50490
    },
    {
      "epoch": 0.41775240931463786,
      "grad_norm": 1148.364990234375,
      "learning_rate": 7.376880323377357e-06,
      "loss": 106.6024,
      "step": 50500
    },
    {
      "epoch": 0.4178351325640071,
      "grad_norm": 567.2054443359375,
      "learning_rate": 7.375639412285877e-06,
      "loss": 94.4765,
      "step": 50510
    },
    {
      "epoch": 0.41791785581337637,
      "grad_norm": 742.42236328125,
      "learning_rate": 7.374398312175674e-06,
      "loss": 103.2163,
      "step": 50520
    },
    {
      "epoch": 0.41800057906274557,
      "grad_norm": 698.6530151367188,
      "learning_rate": 7.373157023145493e-06,
      "loss": 86.0616,
      "step": 50530
    },
    {
      "epoch": 0.4180833023121148,
      "grad_norm": 1061.374755859375,
      "learning_rate": 7.371915545294098e-06,
      "loss": 74.492,
      "step": 50540
    },
    {
      "epoch": 0.4181660255614841,
      "grad_norm": 748.8624877929688,
      "learning_rate": 7.37067387872027e-06,
      "loss": 75.0185,
      "step": 50550
    },
    {
      "epoch": 0.4182487488108533,
      "grad_norm": 315.02374267578125,
      "learning_rate": 7.369432023522801e-06,
      "loss": 77.1597,
      "step": 50560
    },
    {
      "epoch": 0.41833147206022253,
      "grad_norm": 981.3681030273438,
      "learning_rate": 7.3681899798005006e-06,
      "loss": 118.4615,
      "step": 50570
    },
    {
      "epoch": 0.4184141953095918,
      "grad_norm": 1849.541259765625,
      "learning_rate": 7.366947747652191e-06,
      "loss": 104.9723,
      "step": 50580
    },
    {
      "epoch": 0.418496918558961,
      "grad_norm": 1334.4669189453125,
      "learning_rate": 7.365705327176713e-06,
      "loss": 100.1431,
      "step": 50590
    },
    {
      "epoch": 0.41857964180833024,
      "grad_norm": 1951.8896484375,
      "learning_rate": 7.364462718472919e-06,
      "loss": 142.2957,
      "step": 50600
    },
    {
      "epoch": 0.4186623650576995,
      "grad_norm": 659.5794067382812,
      "learning_rate": 7.363219921639677e-06,
      "loss": 96.1194,
      "step": 50610
    },
    {
      "epoch": 0.4187450883070687,
      "grad_norm": 960.1915893554688,
      "learning_rate": 7.361976936775872e-06,
      "loss": 117.1287,
      "step": 50620
    },
    {
      "epoch": 0.41882781155643795,
      "grad_norm": 940.3717651367188,
      "learning_rate": 7.360733763980404e-06,
      "loss": 110.9005,
      "step": 50630
    },
    {
      "epoch": 0.4189105348058072,
      "grad_norm": 741.2780151367188,
      "learning_rate": 7.3594904033521815e-06,
      "loss": 86.0149,
      "step": 50640
    },
    {
      "epoch": 0.4189932580551764,
      "grad_norm": 1676.6136474609375,
      "learning_rate": 7.358246854990138e-06,
      "loss": 119.8522,
      "step": 50650
    },
    {
      "epoch": 0.41907598130454565,
      "grad_norm": 1362.4383544921875,
      "learning_rate": 7.357003118993215e-06,
      "loss": 125.9308,
      "step": 50660
    },
    {
      "epoch": 0.41915870455391485,
      "grad_norm": 1395.822998046875,
      "learning_rate": 7.355759195460371e-06,
      "loss": 125.4457,
      "step": 50670
    },
    {
      "epoch": 0.4192414278032841,
      "grad_norm": 754.3294067382812,
      "learning_rate": 7.354515084490579e-06,
      "loss": 105.6307,
      "step": 50680
    },
    {
      "epoch": 0.41932415105265336,
      "grad_norm": 1079.152587890625,
      "learning_rate": 7.353270786182828e-06,
      "loss": 95.9782,
      "step": 50690
    },
    {
      "epoch": 0.41940687430202256,
      "grad_norm": 713.3341674804688,
      "learning_rate": 7.352026300636121e-06,
      "loss": 104.3826,
      "step": 50700
    },
    {
      "epoch": 0.4194895975513918,
      "grad_norm": 879.9957275390625,
      "learning_rate": 7.350781627949475e-06,
      "loss": 86.2512,
      "step": 50710
    },
    {
      "epoch": 0.41957232080076107,
      "grad_norm": 1013.7462158203125,
      "learning_rate": 7.3495367682219236e-06,
      "loss": 110.0644,
      "step": 50720
    },
    {
      "epoch": 0.41965504405013027,
      "grad_norm": 948.6749877929688,
      "learning_rate": 7.348291721552514e-06,
      "loss": 145.048,
      "step": 50730
    },
    {
      "epoch": 0.4197377672994995,
      "grad_norm": 2777.333251953125,
      "learning_rate": 7.3470464880403105e-06,
      "loss": 116.8931,
      "step": 50740
    },
    {
      "epoch": 0.4198204905488688,
      "grad_norm": 1682.1556396484375,
      "learning_rate": 7.345801067784388e-06,
      "loss": 94.5308,
      "step": 50750
    },
    {
      "epoch": 0.419903213798238,
      "grad_norm": 988.3783569335938,
      "learning_rate": 7.34455546088384e-06,
      "loss": 100.0454,
      "step": 50760
    },
    {
      "epoch": 0.4199859370476072,
      "grad_norm": 721.0875244140625,
      "learning_rate": 7.343309667437775e-06,
      "loss": 116.8833,
      "step": 50770
    },
    {
      "epoch": 0.4200686602969765,
      "grad_norm": 832.2625122070312,
      "learning_rate": 7.3420636875453135e-06,
      "loss": 77.9705,
      "step": 50780
    },
    {
      "epoch": 0.4201513835463457,
      "grad_norm": 995.8257446289062,
      "learning_rate": 7.340817521305595e-06,
      "loss": 101.0271,
      "step": 50790
    },
    {
      "epoch": 0.42023410679571493,
      "grad_norm": 935.89404296875,
      "learning_rate": 7.3395711688177676e-06,
      "loss": 123.7885,
      "step": 50800
    },
    {
      "epoch": 0.4203168300450842,
      "grad_norm": 957.3890380859375,
      "learning_rate": 7.3383246301809985e-06,
      "loss": 79.4234,
      "step": 50810
    },
    {
      "epoch": 0.4203995532944534,
      "grad_norm": 1143.3251953125,
      "learning_rate": 7.337077905494472e-06,
      "loss": 79.35,
      "step": 50820
    },
    {
      "epoch": 0.42048227654382264,
      "grad_norm": 617.7476806640625,
      "learning_rate": 7.335830994857382e-06,
      "loss": 83.5682,
      "step": 50830
    },
    {
      "epoch": 0.4205649997931919,
      "grad_norm": 923.8682861328125,
      "learning_rate": 7.334583898368939e-06,
      "loss": 83.5606,
      "step": 50840
    },
    {
      "epoch": 0.4206477230425611,
      "grad_norm": 456.18902587890625,
      "learning_rate": 7.333336616128369e-06,
      "loss": 110.9043,
      "step": 50850
    },
    {
      "epoch": 0.42073044629193035,
      "grad_norm": 629.4192504882812,
      "learning_rate": 7.332089148234913e-06,
      "loss": 84.0271,
      "step": 50860
    },
    {
      "epoch": 0.4208131695412996,
      "grad_norm": 777.4642333984375,
      "learning_rate": 7.330841494787828e-06,
      "loss": 94.1915,
      "step": 50870
    },
    {
      "epoch": 0.4208958927906688,
      "grad_norm": 1021.780029296875,
      "learning_rate": 7.329593655886382e-06,
      "loss": 86.0737,
      "step": 50880
    },
    {
      "epoch": 0.42097861604003806,
      "grad_norm": 1032.098876953125,
      "learning_rate": 7.3283456316298595e-06,
      "loss": 83.7801,
      "step": 50890
    },
    {
      "epoch": 0.4210613392894073,
      "grad_norm": 987.5896606445312,
      "learning_rate": 7.32709742211756e-06,
      "loss": 86.3313,
      "step": 50900
    },
    {
      "epoch": 0.4211440625387765,
      "grad_norm": 799.8026733398438,
      "learning_rate": 7.325849027448799e-06,
      "loss": 71.7253,
      "step": 50910
    },
    {
      "epoch": 0.42122678578814576,
      "grad_norm": 675.6878051757812,
      "learning_rate": 7.324600447722907e-06,
      "loss": 79.754,
      "step": 50920
    },
    {
      "epoch": 0.421309509037515,
      "grad_norm": 578.5625,
      "learning_rate": 7.323351683039224e-06,
      "loss": 103.9349,
      "step": 50930
    },
    {
      "epoch": 0.4213922322868842,
      "grad_norm": 1150.8740234375,
      "learning_rate": 7.32210273349711e-06,
      "loss": 101.4098,
      "step": 50940
    },
    {
      "epoch": 0.42147495553625347,
      "grad_norm": 1209.810546875,
      "learning_rate": 7.32085359919594e-06,
      "loss": 112.2176,
      "step": 50950
    },
    {
      "epoch": 0.4215576787856227,
      "grad_norm": 1046.1680908203125,
      "learning_rate": 7.3196042802350995e-06,
      "loss": 81.5384,
      "step": 50960
    },
    {
      "epoch": 0.4216404020349919,
      "grad_norm": 1647.843017578125,
      "learning_rate": 7.3183547767139916e-06,
      "loss": 106.6991,
      "step": 50970
    },
    {
      "epoch": 0.4217231252843612,
      "grad_norm": 1031.4130859375,
      "learning_rate": 7.317105088732035e-06,
      "loss": 102.2982,
      "step": 50980
    },
    {
      "epoch": 0.42180584853373043,
      "grad_norm": 739.594482421875,
      "learning_rate": 7.31585521638866e-06,
      "loss": 83.5706,
      "step": 50990
    },
    {
      "epoch": 0.42188857178309963,
      "grad_norm": 1355.9227294921875,
      "learning_rate": 7.314605159783313e-06,
      "loss": 155.9577,
      "step": 51000
    },
    {
      "epoch": 0.4219712950324689,
      "grad_norm": 752.1953735351562,
      "learning_rate": 7.313354919015457e-06,
      "loss": 87.4393,
      "step": 51010
    },
    {
      "epoch": 0.4220540182818381,
      "grad_norm": 1794.4256591796875,
      "learning_rate": 7.312104494184566e-06,
      "loss": 140.2643,
      "step": 51020
    },
    {
      "epoch": 0.42213674153120734,
      "grad_norm": 1220.9268798828125,
      "learning_rate": 7.310853885390133e-06,
      "loss": 98.7091,
      "step": 51030
    },
    {
      "epoch": 0.4222194647805766,
      "grad_norm": 2305.75,
      "learning_rate": 7.309603092731661e-06,
      "loss": 112.4157,
      "step": 51040
    },
    {
      "epoch": 0.4223021880299458,
      "grad_norm": 1095.7064208984375,
      "learning_rate": 7.30835211630867e-06,
      "loss": 102.6219,
      "step": 51050
    },
    {
      "epoch": 0.42238491127931505,
      "grad_norm": 678.8933715820312,
      "learning_rate": 7.3071009562206965e-06,
      "loss": 88.1793,
      "step": 51060
    },
    {
      "epoch": 0.4224676345286843,
      "grad_norm": 809.6482543945312,
      "learning_rate": 7.305849612567287e-06,
      "loss": 99.8888,
      "step": 51070
    },
    {
      "epoch": 0.4225503577780535,
      "grad_norm": 1535.9327392578125,
      "learning_rate": 7.304598085448007e-06,
      "loss": 119.08,
      "step": 51080
    },
    {
      "epoch": 0.42263308102742275,
      "grad_norm": 953.1244506835938,
      "learning_rate": 7.303346374962433e-06,
      "loss": 86.5312,
      "step": 51090
    },
    {
      "epoch": 0.422715804276792,
      "grad_norm": 1713.26318359375,
      "learning_rate": 7.302094481210159e-06,
      "loss": 103.2178,
      "step": 51100
    },
    {
      "epoch": 0.4227985275261612,
      "grad_norm": 826.4808959960938,
      "learning_rate": 7.300842404290792e-06,
      "loss": 96.0054,
      "step": 51110
    },
    {
      "epoch": 0.42288125077553046,
      "grad_norm": 1028.5643310546875,
      "learning_rate": 7.2995901443039554e-06,
      "loss": 128.5591,
      "step": 51120
    },
    {
      "epoch": 0.4229639740248997,
      "grad_norm": 840.199462890625,
      "learning_rate": 7.298337701349285e-06,
      "loss": 116.7384,
      "step": 51130
    },
    {
      "epoch": 0.4230466972742689,
      "grad_norm": 672.873779296875,
      "learning_rate": 7.29708507552643e-06,
      "loss": 68.3276,
      "step": 51140
    },
    {
      "epoch": 0.42312942052363817,
      "grad_norm": 383.77313232421875,
      "learning_rate": 7.295832266935059e-06,
      "loss": 84.0062,
      "step": 51150
    },
    {
      "epoch": 0.4232121437730074,
      "grad_norm": 1001.8519287109375,
      "learning_rate": 7.2945792756748505e-06,
      "loss": 140.7101,
      "step": 51160
    },
    {
      "epoch": 0.4232948670223766,
      "grad_norm": 709.9981689453125,
      "learning_rate": 7.2933261018455005e-06,
      "loss": 75.1831,
      "step": 51170
    },
    {
      "epoch": 0.4233775902717459,
      "grad_norm": 675.74658203125,
      "learning_rate": 7.292072745546716e-06,
      "loss": 73.5509,
      "step": 51180
    },
    {
      "epoch": 0.42346031352111513,
      "grad_norm": 886.9816284179688,
      "learning_rate": 7.290819206878223e-06,
      "loss": 92.8137,
      "step": 51190
    },
    {
      "epoch": 0.4235430367704843,
      "grad_norm": 688.6138305664062,
      "learning_rate": 7.289565485939759e-06,
      "loss": 109.6997,
      "step": 51200
    },
    {
      "epoch": 0.4236257600198536,
      "grad_norm": 1372.497802734375,
      "learning_rate": 7.288311582831078e-06,
      "loss": 86.5049,
      "step": 51210
    },
    {
      "epoch": 0.42370848326922284,
      "grad_norm": 632.5243530273438,
      "learning_rate": 7.2870574976519455e-06,
      "loss": 73.3826,
      "step": 51220
    },
    {
      "epoch": 0.42379120651859203,
      "grad_norm": 564.6524658203125,
      "learning_rate": 7.2858032305021455e-06,
      "loss": 123.0075,
      "step": 51230
    },
    {
      "epoch": 0.4238739297679613,
      "grad_norm": 758.579833984375,
      "learning_rate": 7.28454878148147e-06,
      "loss": 97.95,
      "step": 51240
    },
    {
      "epoch": 0.42395665301733054,
      "grad_norm": 1073.8216552734375,
      "learning_rate": 7.283294150689735e-06,
      "loss": 88.5506,
      "step": 51250
    },
    {
      "epoch": 0.42403937626669974,
      "grad_norm": 621.6570434570312,
      "learning_rate": 7.282039338226763e-06,
      "loss": 106.7351,
      "step": 51260
    },
    {
      "epoch": 0.424122099516069,
      "grad_norm": 1591.994384765625,
      "learning_rate": 7.280784344192393e-06,
      "loss": 117.9606,
      "step": 51270
    },
    {
      "epoch": 0.42420482276543825,
      "grad_norm": 811.947265625,
      "learning_rate": 7.279529168686481e-06,
      "loss": 83.5623,
      "step": 51280
    },
    {
      "epoch": 0.42428754601480745,
      "grad_norm": 939.7516479492188,
      "learning_rate": 7.278273811808894e-06,
      "loss": 91.11,
      "step": 51290
    },
    {
      "epoch": 0.4243702692641767,
      "grad_norm": 1210.847900390625,
      "learning_rate": 7.2770182736595164e-06,
      "loss": 130.9754,
      "step": 51300
    },
    {
      "epoch": 0.42445299251354596,
      "grad_norm": 1220.7769775390625,
      "learning_rate": 7.275762554338244e-06,
      "loss": 76.7486,
      "step": 51310
    },
    {
      "epoch": 0.42453571576291516,
      "grad_norm": 1265.53271484375,
      "learning_rate": 7.2745066539449905e-06,
      "loss": 95.9141,
      "step": 51320
    },
    {
      "epoch": 0.4246184390122844,
      "grad_norm": 875.6475219726562,
      "learning_rate": 7.27325057257968e-06,
      "loss": 97.6407,
      "step": 51330
    },
    {
      "epoch": 0.42470116226165366,
      "grad_norm": 693.2019653320312,
      "learning_rate": 7.271994310342254e-06,
      "loss": 99.6023,
      "step": 51340
    },
    {
      "epoch": 0.42478388551102286,
      "grad_norm": 897.9375,
      "learning_rate": 7.270737867332669e-06,
      "loss": 104.1033,
      "step": 51350
    },
    {
      "epoch": 0.4248666087603921,
      "grad_norm": 1381.2633056640625,
      "learning_rate": 7.2694812436508934e-06,
      "loss": 98.0639,
      "step": 51360
    },
    {
      "epoch": 0.4249493320097613,
      "grad_norm": 825.0390625,
      "learning_rate": 7.268224439396909e-06,
      "loss": 109.9973,
      "step": 51370
    },
    {
      "epoch": 0.42503205525913057,
      "grad_norm": 997.9774780273438,
      "learning_rate": 7.266967454670717e-06,
      "loss": 115.4315,
      "step": 51380
    },
    {
      "epoch": 0.4251147785084998,
      "grad_norm": 1014.7069702148438,
      "learning_rate": 7.265710289572328e-06,
      "loss": 122.6427,
      "step": 51390
    },
    {
      "epoch": 0.425197501757869,
      "grad_norm": 1186.5941162109375,
      "learning_rate": 7.264452944201771e-06,
      "loss": 79.164,
      "step": 51400
    },
    {
      "epoch": 0.4252802250072383,
      "grad_norm": 896.6864013671875,
      "learning_rate": 7.263195418659083e-06,
      "loss": 94.9668,
      "step": 51410
    },
    {
      "epoch": 0.42536294825660753,
      "grad_norm": 1269.0845947265625,
      "learning_rate": 7.261937713044325e-06,
      "loss": 111.4878,
      "step": 51420
    },
    {
      "epoch": 0.42544567150597673,
      "grad_norm": 737.5647583007812,
      "learning_rate": 7.260679827457562e-06,
      "loss": 144.4097,
      "step": 51430
    },
    {
      "epoch": 0.425528394755346,
      "grad_norm": 1106.5899658203125,
      "learning_rate": 7.259421761998881e-06,
      "loss": 97.2455,
      "step": 51440
    },
    {
      "epoch": 0.42561111800471524,
      "grad_norm": 867.489013671875,
      "learning_rate": 7.2581635167683805e-06,
      "loss": 105.3224,
      "step": 51450
    },
    {
      "epoch": 0.42569384125408444,
      "grad_norm": 693.6383666992188,
      "learning_rate": 7.256905091866171e-06,
      "loss": 101.9842,
      "step": 51460
    },
    {
      "epoch": 0.4257765645034537,
      "grad_norm": 611.8115844726562,
      "learning_rate": 7.255646487392382e-06,
      "loss": 115.6671,
      "step": 51470
    },
    {
      "epoch": 0.42585928775282295,
      "grad_norm": 1504.69775390625,
      "learning_rate": 7.254387703447154e-06,
      "loss": 95.3933,
      "step": 51480
    },
    {
      "epoch": 0.42594201100219214,
      "grad_norm": 600.2556762695312,
      "learning_rate": 7.2531287401306435e-06,
      "loss": 92.5239,
      "step": 51490
    },
    {
      "epoch": 0.4260247342515614,
      "grad_norm": 407.8628845214844,
      "learning_rate": 7.251869597543019e-06,
      "loss": 82.7194,
      "step": 51500
    },
    {
      "epoch": 0.42610745750093065,
      "grad_norm": 614.1116943359375,
      "learning_rate": 7.250610275784464e-06,
      "loss": 93.6232,
      "step": 51510
    },
    {
      "epoch": 0.42619018075029985,
      "grad_norm": 898.9727172851562,
      "learning_rate": 7.2493507749551795e-06,
      "loss": 85.4422,
      "step": 51520
    },
    {
      "epoch": 0.4262729039996691,
      "grad_norm": 1155.338623046875,
      "learning_rate": 7.248091095155378e-06,
      "loss": 118.6169,
      "step": 51530
    },
    {
      "epoch": 0.42635562724903836,
      "grad_norm": 1159.13916015625,
      "learning_rate": 7.246831236485283e-06,
      "loss": 135.8041,
      "step": 51540
    },
    {
      "epoch": 0.42643835049840756,
      "grad_norm": 928.2463989257812,
      "learning_rate": 7.245571199045139e-06,
      "loss": 108.4106,
      "step": 51550
    },
    {
      "epoch": 0.4265210737477768,
      "grad_norm": 767.41259765625,
      "learning_rate": 7.244310982935202e-06,
      "loss": 74.6865,
      "step": 51560
    },
    {
      "epoch": 0.42660379699714607,
      "grad_norm": 709.4650268554688,
      "learning_rate": 7.243050588255738e-06,
      "loss": 76.4944,
      "step": 51570
    },
    {
      "epoch": 0.42668652024651527,
      "grad_norm": 1164.441162109375,
      "learning_rate": 7.241790015107034e-06,
      "loss": 88.2155,
      "step": 51580
    },
    {
      "epoch": 0.4267692434958845,
      "grad_norm": 1503.4273681640625,
      "learning_rate": 7.240529263589386e-06,
      "loss": 126.4441,
      "step": 51590
    },
    {
      "epoch": 0.4268519667452538,
      "grad_norm": 760.6532592773438,
      "learning_rate": 7.239268333803109e-06,
      "loss": 105.8668,
      "step": 51600
    },
    {
      "epoch": 0.426934689994623,
      "grad_norm": 1101.9364013671875,
      "learning_rate": 7.2380072258485265e-06,
      "loss": 92.2242,
      "step": 51610
    },
    {
      "epoch": 0.4270174132439922,
      "grad_norm": 554.7161254882812,
      "learning_rate": 7.2367459398259795e-06,
      "loss": 83.6779,
      "step": 51620
    },
    {
      "epoch": 0.4271001364933615,
      "grad_norm": 972.8421020507812,
      "learning_rate": 7.2354844758358234e-06,
      "loss": 123.7181,
      "step": 51630
    },
    {
      "epoch": 0.4271828597427307,
      "grad_norm": 1161.73828125,
      "learning_rate": 7.234222833978427e-06,
      "loss": 106.6433,
      "step": 51640
    },
    {
      "epoch": 0.42726558299209993,
      "grad_norm": 1067.09765625,
      "learning_rate": 7.232961014354175e-06,
      "loss": 104.9934,
      "step": 51650
    },
    {
      "epoch": 0.4273483062414692,
      "grad_norm": 1101.327392578125,
      "learning_rate": 7.23169901706346e-06,
      "loss": 106.5562,
      "step": 51660
    },
    {
      "epoch": 0.4274310294908384,
      "grad_norm": 1341.46044921875,
      "learning_rate": 7.2304368422067e-06,
      "loss": 134.7537,
      "step": 51670
    },
    {
      "epoch": 0.42751375274020764,
      "grad_norm": 1436.9295654296875,
      "learning_rate": 7.2291744898843145e-06,
      "loss": 98.4549,
      "step": 51680
    },
    {
      "epoch": 0.4275964759895769,
      "grad_norm": 540.67236328125,
      "learning_rate": 7.227911960196746e-06,
      "loss": 86.389,
      "step": 51690
    },
    {
      "epoch": 0.4276791992389461,
      "grad_norm": 1108.2845458984375,
      "learning_rate": 7.226649253244448e-06,
      "loss": 74.8376,
      "step": 51700
    },
    {
      "epoch": 0.42776192248831535,
      "grad_norm": 964.5218505859375,
      "learning_rate": 7.225386369127886e-06,
      "loss": 112.0215,
      "step": 51710
    },
    {
      "epoch": 0.4278446457376846,
      "grad_norm": 768.393310546875,
      "learning_rate": 7.224123307947545e-06,
      "loss": 94.5367,
      "step": 51720
    },
    {
      "epoch": 0.4279273689870538,
      "grad_norm": 708.0656127929688,
      "learning_rate": 7.2228600698039205e-06,
      "loss": 108.4423,
      "step": 51730
    },
    {
      "epoch": 0.42801009223642306,
      "grad_norm": 1345.224853515625,
      "learning_rate": 7.221596654797522e-06,
      "loss": 91.7173,
      "step": 51740
    },
    {
      "epoch": 0.42809281548579226,
      "grad_norm": 1059.3037109375,
      "learning_rate": 7.2203330630288714e-06,
      "loss": 109.0566,
      "step": 51750
    },
    {
      "epoch": 0.4281755387351615,
      "grad_norm": 975.2772216796875,
      "learning_rate": 7.21906929459851e-06,
      "loss": 140.6817,
      "step": 51760
    },
    {
      "epoch": 0.42825826198453076,
      "grad_norm": 1177.9395751953125,
      "learning_rate": 7.217805349606988e-06,
      "loss": 91.187,
      "step": 51770
    },
    {
      "epoch": 0.42834098523389996,
      "grad_norm": 685.8370971679688,
      "learning_rate": 7.216541228154875e-06,
      "loss": 93.8095,
      "step": 51780
    },
    {
      "epoch": 0.4284237084832692,
      "grad_norm": 1138.31689453125,
      "learning_rate": 7.215276930342747e-06,
      "loss": 104.4566,
      "step": 51790
    },
    {
      "epoch": 0.42850643173263847,
      "grad_norm": 601.1682739257812,
      "learning_rate": 7.214012456271202e-06,
      "loss": 130.8603,
      "step": 51800
    },
    {
      "epoch": 0.42858915498200767,
      "grad_norm": 1000.6312866210938,
      "learning_rate": 7.212747806040845e-06,
      "loss": 123.3161,
      "step": 51810
    },
    {
      "epoch": 0.4286718782313769,
      "grad_norm": 1142.5823974609375,
      "learning_rate": 7.211482979752302e-06,
      "loss": 85.4368,
      "step": 51820
    },
    {
      "epoch": 0.4287546014807462,
      "grad_norm": 816.385986328125,
      "learning_rate": 7.210217977506207e-06,
      "loss": 109.9975,
      "step": 51830
    },
    {
      "epoch": 0.4288373247301154,
      "grad_norm": 944.0142822265625,
      "learning_rate": 7.208952799403211e-06,
      "loss": 108.7334,
      "step": 51840
    },
    {
      "epoch": 0.42892004797948463,
      "grad_norm": 824.8801879882812,
      "learning_rate": 7.207687445543977e-06,
      "loss": 82.7929,
      "step": 51850
    },
    {
      "epoch": 0.4290027712288539,
      "grad_norm": 964.5783081054688,
      "learning_rate": 7.206421916029187e-06,
      "loss": 121.5998,
      "step": 51860
    },
    {
      "epoch": 0.4290854944782231,
      "grad_norm": 1063.656982421875,
      "learning_rate": 7.205156210959529e-06,
      "loss": 90.583,
      "step": 51870
    },
    {
      "epoch": 0.42916821772759234,
      "grad_norm": 3292.513671875,
      "learning_rate": 7.203890330435715e-06,
      "loss": 105.8095,
      "step": 51880
    },
    {
      "epoch": 0.4292509409769616,
      "grad_norm": 703.033447265625,
      "learning_rate": 7.202624274558458e-06,
      "loss": 106.7044,
      "step": 51890
    },
    {
      "epoch": 0.4293336642263308,
      "grad_norm": 669.6672973632812,
      "learning_rate": 7.201358043428499e-06,
      "loss": 89.5573,
      "step": 51900
    },
    {
      "epoch": 0.42941638747570005,
      "grad_norm": 647.8473510742188,
      "learning_rate": 7.200091637146582e-06,
      "loss": 99.5425,
      "step": 51910
    },
    {
      "epoch": 0.4294991107250693,
      "grad_norm": 1656.5313720703125,
      "learning_rate": 7.198825055813471e-06,
      "loss": 120.4202,
      "step": 51920
    },
    {
      "epoch": 0.4295818339744385,
      "grad_norm": 755.7421875,
      "learning_rate": 7.197558299529941e-06,
      "loss": 74.5723,
      "step": 51930
    },
    {
      "epoch": 0.42966455722380775,
      "grad_norm": 1194.8021240234375,
      "learning_rate": 7.196291368396784e-06,
      "loss": 104.8849,
      "step": 51940
    },
    {
      "epoch": 0.429747280473177,
      "grad_norm": 810.8287353515625,
      "learning_rate": 7.1950242625148e-06,
      "loss": 109.7433,
      "step": 51950
    },
    {
      "epoch": 0.4298300037225462,
      "grad_norm": 5541.013671875,
      "learning_rate": 7.1937569819848115e-06,
      "loss": 117.9274,
      "step": 51960
    },
    {
      "epoch": 0.42991272697191546,
      "grad_norm": 1193.3897705078125,
      "learning_rate": 7.192489526907646e-06,
      "loss": 112.9372,
      "step": 51970
    },
    {
      "epoch": 0.4299954502212847,
      "grad_norm": 1110.0302734375,
      "learning_rate": 7.191221897384153e-06,
      "loss": 120.9369,
      "step": 51980
    },
    {
      "epoch": 0.4300781734706539,
      "grad_norm": 1447.90771484375,
      "learning_rate": 7.189954093515189e-06,
      "loss": 142.4958,
      "step": 51990
    },
    {
      "epoch": 0.43016089672002317,
      "grad_norm": 1462.6444091796875,
      "learning_rate": 7.188686115401628e-06,
      "loss": 127.5024,
      "step": 52000
    }
  ],
  "logging_steps": 10,
  "max_steps": 123750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}