{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1858450698535056,
  "eval_steps": 500,
  "global_step": 23000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.080220428413287e-05,
      "grad_norm": 120784.5234375,
      "learning_rate": 4.040404040404041e-08,
      "loss": 6344.0191,
      "step": 10
    },
    {
      "epoch": 0.00016160440856826573,
      "grad_norm": 246101.4375,
      "learning_rate": 8.080808080808082e-08,
      "loss": 7230.2391,
      "step": 20
    },
    {
      "epoch": 0.00024240661285239863,
      "grad_norm": 220140.828125,
      "learning_rate": 1.2121212121212122e-07,
      "loss": 7248.1844,
      "step": 30
    },
    {
      "epoch": 0.00032320881713653147,
      "grad_norm": 468224.40625,
      "learning_rate": 1.6161616161616163e-07,
      "loss": 8245.5844,
      "step": 40
    },
    {
      "epoch": 0.00040401102142066436,
      "grad_norm": 129730.1171875,
      "learning_rate": 2.0202020202020202e-07,
      "loss": 5464.3164,
      "step": 50
    },
    {
      "epoch": 0.00048481322570479725,
      "grad_norm": 37984.21484375,
      "learning_rate": 2.4242424242424244e-07,
      "loss": 7551.0562,
      "step": 60
    },
    {
      "epoch": 0.0005656154299889301,
      "grad_norm": 170187.078125,
      "learning_rate": 2.8282828282828283e-07,
      "loss": 6856.2141,
      "step": 70
    },
    {
      "epoch": 0.0006464176342730629,
      "grad_norm": 64738.05078125,
      "learning_rate": 3.2323232323232327e-07,
      "loss": 7045.2477,
      "step": 80
    },
    {
      "epoch": 0.0007272198385571959,
      "grad_norm": 114463.8515625,
      "learning_rate": 3.6363636363636366e-07,
      "loss": 5803.459,
      "step": 90
    },
    {
      "epoch": 0.0008080220428413287,
      "grad_norm": 102587.6796875,
      "learning_rate": 4.0404040404040405e-07,
      "loss": 3976.1211,
      "step": 100
    },
    {
      "epoch": 0.0008888242471254616,
      "grad_norm": 78112.53125,
      "learning_rate": 4.444444444444445e-07,
      "loss": 3535.432,
      "step": 110
    },
    {
      "epoch": 0.0009696264514095945,
      "grad_norm": 100944.5703125,
      "learning_rate": 4.848484848484849e-07,
      "loss": 4244.4645,
      "step": 120
    },
    {
      "epoch": 0.0010504286556937273,
      "grad_norm": 178465.359375,
      "learning_rate": 5.252525252525253e-07,
      "loss": 5655.2863,
      "step": 130
    },
    {
      "epoch": 0.0011312308599778602,
      "grad_norm": 70585.359375,
      "learning_rate": 5.656565656565657e-07,
      "loss": 3637.5656,
      "step": 140
    },
    {
      "epoch": 0.001212033064261993,
      "grad_norm": 51223.41015625,
      "learning_rate": 6.060606060606061e-07,
      "loss": 2629.8635,
      "step": 150
    },
    {
      "epoch": 0.0012928352685461259,
      "grad_norm": 53755.01953125,
      "learning_rate": 6.464646464646465e-07,
      "loss": 4522.2559,
      "step": 160
    },
    {
      "epoch": 0.0013736374728302587,
      "grad_norm": 72124.0625,
      "learning_rate": 6.868686868686869e-07,
      "loss": 3395.8777,
      "step": 170
    },
    {
      "epoch": 0.0014544396771143918,
      "grad_norm": 11780.193359375,
      "learning_rate": 7.272727272727273e-07,
      "loss": 1784.9486,
      "step": 180
    },
    {
      "epoch": 0.0015352418813985246,
      "grad_norm": 62803.8828125,
      "learning_rate": 7.676767676767678e-07,
      "loss": 2220.277,
      "step": 190
    },
    {
      "epoch": 0.0016160440856826574,
      "grad_norm": 118829.34375,
      "learning_rate": 8.080808080808081e-07,
      "loss": 2133.5133,
      "step": 200
    },
    {
      "epoch": 0.0016968462899667903,
      "grad_norm": 7391.849609375,
      "learning_rate": 8.484848484848486e-07,
      "loss": 1618.4662,
      "step": 210
    },
    {
      "epoch": 0.0017776484942509231,
      "grad_norm": 7889.263671875,
      "learning_rate": 8.88888888888889e-07,
      "loss": 1578.3401,
      "step": 220
    },
    {
      "epoch": 0.001858450698535056,
      "grad_norm": 10828.140625,
      "learning_rate": 9.292929292929294e-07,
      "loss": 1327.2748,
      "step": 230
    },
    {
      "epoch": 0.001939252902819189,
      "grad_norm": 15991.2119140625,
      "learning_rate": 9.696969696969698e-07,
      "loss": 1342.3516,
      "step": 240
    },
    {
      "epoch": 0.0020200551071033216,
      "grad_norm": 7872.46484375,
      "learning_rate": 1.0101010101010103e-06,
      "loss": 1077.598,
      "step": 250
    },
    {
      "epoch": 0.0021008573113874547,
      "grad_norm": 3977.63623046875,
      "learning_rate": 1.0505050505050506e-06,
      "loss": 835.4583,
      "step": 260
    },
    {
      "epoch": 0.0021816595156715873,
      "grad_norm": 3596.71923828125,
      "learning_rate": 1.090909090909091e-06,
      "loss": 766.0804,
      "step": 270
    },
    {
      "epoch": 0.0022624617199557204,
      "grad_norm": 10194.791015625,
      "learning_rate": 1.1313131313131313e-06,
      "loss": 777.5695,
      "step": 280
    },
    {
      "epoch": 0.0023432639242398534,
      "grad_norm": 7022.6103515625,
      "learning_rate": 1.1717171717171719e-06,
      "loss": 639.281,
      "step": 290
    },
    {
      "epoch": 0.002424066128523986,
      "grad_norm": 2920.81787109375,
      "learning_rate": 1.2121212121212122e-06,
      "loss": 703.318,
      "step": 300
    },
    {
      "epoch": 0.002504868332808119,
      "grad_norm": 4137.4970703125,
      "learning_rate": 1.2525252525252527e-06,
      "loss": 766.8718,
      "step": 310
    },
    {
      "epoch": 0.0025856705370922517,
      "grad_norm": 8379.064453125,
      "learning_rate": 1.292929292929293e-06,
      "loss": 625.7259,
      "step": 320
    },
    {
      "epoch": 0.002666472741376385,
      "grad_norm": 2632.939697265625,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 529.0049,
      "step": 330
    },
    {
      "epoch": 0.0027472749456605174,
      "grad_norm": 3072.3486328125,
      "learning_rate": 1.3737373737373738e-06,
      "loss": 557.5195,
      "step": 340
    },
    {
      "epoch": 0.0028280771499446505,
      "grad_norm": 2430.994384765625,
      "learning_rate": 1.4141414141414143e-06,
      "loss": 548.776,
      "step": 350
    },
    {
      "epoch": 0.0029088793542287835,
      "grad_norm": 2531.90771484375,
      "learning_rate": 1.4545454545454546e-06,
      "loss": 602.2291,
      "step": 360
    },
    {
      "epoch": 0.002989681558512916,
      "grad_norm": 5374.16552734375,
      "learning_rate": 1.4949494949494952e-06,
      "loss": 447.6404,
      "step": 370
    },
    {
      "epoch": 0.003070483762797049,
      "grad_norm": 1357.3726806640625,
      "learning_rate": 1.5353535353535355e-06,
      "loss": 342.7574,
      "step": 380
    },
    {
      "epoch": 0.003151285967081182,
      "grad_norm": 1249.0936279296875,
      "learning_rate": 1.5757575757575759e-06,
      "loss": 562.0835,
      "step": 390
    },
    {
      "epoch": 0.003232088171365315,
      "grad_norm": 1270.3609619140625,
      "learning_rate": 1.6161616161616162e-06,
      "loss": 467.299,
      "step": 400
    },
    {
      "epoch": 0.0033128903756494475,
      "grad_norm": 2576.8291015625,
      "learning_rate": 1.6565656565656565e-06,
      "loss": 448.897,
      "step": 410
    },
    {
      "epoch": 0.0033936925799335806,
      "grad_norm": 1376.2908935546875,
      "learning_rate": 1.6969696969696973e-06,
      "loss": 382.242,
      "step": 420
    },
    {
      "epoch": 0.0034744947842177136,
      "grad_norm": 1212.802490234375,
      "learning_rate": 1.7373737373737376e-06,
      "loss": 486.5346,
      "step": 430
    },
    {
      "epoch": 0.0035552969885018462,
      "grad_norm": 801.7883911132812,
      "learning_rate": 1.777777777777778e-06,
      "loss": 452.1537,
      "step": 440
    },
    {
      "epoch": 0.0036360991927859793,
      "grad_norm": 1590.736572265625,
      "learning_rate": 1.818181818181818e-06,
      "loss": 449.5157,
      "step": 450
    },
    {
      "epoch": 0.003716901397070112,
      "grad_norm": 3368.974853515625,
      "learning_rate": 1.8585858585858588e-06,
      "loss": 445.9489,
      "step": 460
    },
    {
      "epoch": 0.003797703601354245,
      "grad_norm": 2164.530517578125,
      "learning_rate": 1.8989898989898992e-06,
      "loss": 449.8674,
      "step": 470
    },
    {
      "epoch": 0.003878505805638378,
      "grad_norm": 1234.2860107421875,
      "learning_rate": 1.9393939393939395e-06,
      "loss": 375.5616,
      "step": 480
    },
    {
      "epoch": 0.003959308009922511,
      "grad_norm": 1191.9912109375,
      "learning_rate": 1.9797979797979796e-06,
      "loss": 394.9626,
      "step": 490
    },
    {
      "epoch": 0.004040110214206643,
      "grad_norm": 2829.901611328125,
      "learning_rate": 2.0202020202020206e-06,
      "loss": 398.194,
      "step": 500
    },
    {
      "epoch": 0.004120912418490777,
      "grad_norm": 3573.80322265625,
      "learning_rate": 2.0606060606060607e-06,
      "loss": 440.5374,
      "step": 510
    },
    {
      "epoch": 0.004201714622774909,
      "grad_norm": 991.9559326171875,
      "learning_rate": 2.1010101010101013e-06,
      "loss": 552.4083,
      "step": 520
    },
    {
      "epoch": 0.004282516827059042,
      "grad_norm": 2654.274658203125,
      "learning_rate": 2.1414141414141414e-06,
      "loss": 414.1354,
      "step": 530
    },
    {
      "epoch": 0.004363319031343175,
      "grad_norm": 1416.258056640625,
      "learning_rate": 2.181818181818182e-06,
      "loss": 443.3496,
      "step": 540
    },
    {
      "epoch": 0.004444121235627308,
      "grad_norm": 1433.0880126953125,
      "learning_rate": 2.2222222222222225e-06,
      "loss": 476.5702,
      "step": 550
    },
    {
      "epoch": 0.004524923439911441,
      "grad_norm": 852.6603393554688,
      "learning_rate": 2.2626262626262626e-06,
      "loss": 425.1744,
      "step": 560
    },
    {
      "epoch": 0.004605725644195573,
      "grad_norm": 1662.8759765625,
      "learning_rate": 2.303030303030303e-06,
      "loss": 432.2859,
      "step": 570
    },
    {
      "epoch": 0.004686527848479707,
      "grad_norm": 1177.7154541015625,
      "learning_rate": 2.3434343434343437e-06,
      "loss": 472.836,
      "step": 580
    },
    {
      "epoch": 0.0047673300527638395,
      "grad_norm": 1203.2510986328125,
      "learning_rate": 2.383838383838384e-06,
      "loss": 400.3267,
      "step": 590
    },
    {
      "epoch": 0.004848132257047972,
      "grad_norm": 806.4482421875,
      "learning_rate": 2.4242424242424244e-06,
      "loss": 381.2281,
      "step": 600
    },
    {
      "epoch": 0.004928934461332105,
      "grad_norm": 1223.798095703125,
      "learning_rate": 2.4646464646464645e-06,
      "loss": 314.522,
      "step": 610
    },
    {
      "epoch": 0.005009736665616238,
      "grad_norm": 1220.419189453125,
      "learning_rate": 2.5050505050505055e-06,
      "loss": 391.7079,
      "step": 620
    },
    {
      "epoch": 0.005090538869900371,
      "grad_norm": 1846.539306640625,
      "learning_rate": 2.5454545454545456e-06,
      "loss": 415.6819,
      "step": 630
    },
    {
      "epoch": 0.0051713410741845035,
      "grad_norm": 1243.7620849609375,
      "learning_rate": 2.585858585858586e-06,
      "loss": 437.1002,
      "step": 640
    },
    {
      "epoch": 0.005252143278468637,
      "grad_norm": 1140.9453125,
      "learning_rate": 2.6262626262626263e-06,
      "loss": 415.1159,
      "step": 650
    },
    {
      "epoch": 0.00533294548275277,
      "grad_norm": 4821.37060546875,
      "learning_rate": 2.666666666666667e-06,
      "loss": 464.5417,
      "step": 660
    },
    {
      "epoch": 0.005413747687036902,
      "grad_norm": 1713.6087646484375,
      "learning_rate": 2.7070707070707074e-06,
      "loss": 375.5785,
      "step": 670
    },
    {
      "epoch": 0.005494549891321035,
      "grad_norm": 2620.644287109375,
      "learning_rate": 2.7474747474747475e-06,
      "loss": 299.2374,
      "step": 680
    },
    {
      "epoch": 0.005575352095605168,
      "grad_norm": 1759.0950927734375,
      "learning_rate": 2.787878787878788e-06,
      "loss": 415.4854,
      "step": 690
    },
    {
      "epoch": 0.005656154299889301,
      "grad_norm": 1536.8719482421875,
      "learning_rate": 2.8282828282828286e-06,
      "loss": 449.5622,
      "step": 700
    },
    {
      "epoch": 0.0057369565041734336,
      "grad_norm": 2095.785400390625,
      "learning_rate": 2.8686868686868687e-06,
      "loss": 380.9472,
      "step": 710
    },
    {
      "epoch": 0.005817758708457567,
      "grad_norm": 1478.4825439453125,
      "learning_rate": 2.9090909090909093e-06,
      "loss": 390.7729,
      "step": 720
    },
    {
      "epoch": 0.0058985609127417,
      "grad_norm": 1876.679443359375,
      "learning_rate": 2.9494949494949494e-06,
      "loss": 391.9601,
      "step": 730
    },
    {
      "epoch": 0.005979363117025832,
      "grad_norm": 749.0634765625,
      "learning_rate": 2.9898989898989904e-06,
      "loss": 253.9756,
      "step": 740
    },
    {
      "epoch": 0.006060165321309965,
      "grad_norm": 906.431396484375,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 347.2193,
      "step": 750
    },
    {
      "epoch": 0.006140967525594098,
      "grad_norm": 1034.1180419921875,
      "learning_rate": 3.070707070707071e-06,
      "loss": 311.8948,
      "step": 760
    },
    {
      "epoch": 0.006221769729878231,
      "grad_norm": 2706.747802734375,
      "learning_rate": 3.111111111111111e-06,
      "loss": 329.9375,
      "step": 770
    },
    {
      "epoch": 0.006302571934162364,
      "grad_norm": 1195.091064453125,
      "learning_rate": 3.1515151515151517e-06,
      "loss": 303.2004,
      "step": 780
    },
    {
      "epoch": 0.006383374138446497,
      "grad_norm": 1622.099609375,
      "learning_rate": 3.191919191919192e-06,
      "loss": 359.3405,
      "step": 790
    },
    {
      "epoch": 0.00646417634273063,
      "grad_norm": 1255.0582275390625,
      "learning_rate": 3.2323232323232324e-06,
      "loss": 362.8006,
      "step": 800
    },
    {
      "epoch": 0.006544978547014762,
      "grad_norm": 1571.966552734375,
      "learning_rate": 3.2727272727272733e-06,
      "loss": 372.003,
      "step": 810
    },
    {
      "epoch": 0.006625780751298895,
      "grad_norm": 1350.624755859375,
      "learning_rate": 3.313131313131313e-06,
      "loss": 391.2471,
      "step": 820
    },
    {
      "epoch": 0.0067065829555830285,
      "grad_norm": 1288.0430908203125,
      "learning_rate": 3.3535353535353536e-06,
      "loss": 381.0373,
      "step": 830
    },
    {
      "epoch": 0.006787385159867161,
      "grad_norm": 1756.347900390625,
      "learning_rate": 3.3939393939393946e-06,
      "loss": 366.8195,
      "step": 840
    },
    {
      "epoch": 0.006868187364151294,
      "grad_norm": 1164.05224609375,
      "learning_rate": 3.4343434343434343e-06,
      "loss": 340.2402,
      "step": 850
    },
    {
      "epoch": 0.006948989568435427,
      "grad_norm": 6004.3291015625,
      "learning_rate": 3.4747474747474752e-06,
      "loss": 397.7737,
      "step": 860
    },
    {
      "epoch": 0.00702979177271956,
      "grad_norm": 1949.6370849609375,
      "learning_rate": 3.515151515151515e-06,
      "loss": 352.9498,
      "step": 870
    },
    {
      "epoch": 0.0071105939770036925,
      "grad_norm": 823.8360595703125,
      "learning_rate": 3.555555555555556e-06,
      "loss": 432.1595,
      "step": 880
    },
    {
      "epoch": 0.007191396181287825,
      "grad_norm": 3512.72607421875,
      "learning_rate": 3.5959595959595965e-06,
      "loss": 342.5901,
      "step": 890
    },
    {
      "epoch": 0.007272198385571959,
      "grad_norm": 1352.2506103515625,
      "learning_rate": 3.636363636363636e-06,
      "loss": 306.887,
      "step": 900
    },
    {
      "epoch": 0.007353000589856091,
      "grad_norm": 2983.867919921875,
      "learning_rate": 3.676767676767677e-06,
      "loss": 381.7238,
      "step": 910
    },
    {
      "epoch": 0.007433802794140224,
      "grad_norm": 2423.1806640625,
      "learning_rate": 3.7171717171717177e-06,
      "loss": 303.3808,
      "step": 920
    },
    {
      "epoch": 0.007514604998424357,
      "grad_norm": 953.1580810546875,
      "learning_rate": 3.757575757575758e-06,
      "loss": 380.6255,
      "step": 930
    },
    {
      "epoch": 0.00759540720270849,
      "grad_norm": 5818.9609375,
      "learning_rate": 3.7979797979797984e-06,
      "loss": 372.4149,
      "step": 940
    },
    {
      "epoch": 0.007676209406992623,
      "grad_norm": 1317.5467529296875,
      "learning_rate": 3.8383838383838385e-06,
      "loss": 369.7815,
      "step": 950
    },
    {
      "epoch": 0.007757011611276756,
      "grad_norm": 929.1298217773438,
      "learning_rate": 3.878787878787879e-06,
      "loss": 395.9197,
      "step": 960
    },
    {
      "epoch": 0.007837813815560889,
      "grad_norm": 1342.6861572265625,
      "learning_rate": 3.9191919191919196e-06,
      "loss": 384.7229,
      "step": 970
    },
    {
      "epoch": 0.007918616019845021,
      "grad_norm": 1253.3720703125,
      "learning_rate": 3.959595959595959e-06,
      "loss": 302.0778,
      "step": 980
    },
    {
      "epoch": 0.007999418224129154,
      "grad_norm": 966.1517333984375,
      "learning_rate": 4.000000000000001e-06,
      "loss": 407.2742,
      "step": 990
    },
    {
      "epoch": 0.008080220428413287,
      "grad_norm": 949.992919921875,
      "learning_rate": 4.040404040404041e-06,
      "loss": 333.9852,
      "step": 1000
    },
    {
      "epoch": 0.00816102263269742,
      "grad_norm": 2160.320068359375,
      "learning_rate": 4.080808080808081e-06,
      "loss": 320.1485,
      "step": 1010
    },
    {
      "epoch": 0.008241824836981554,
      "grad_norm": 1784.5238037109375,
      "learning_rate": 4.1212121212121215e-06,
      "loss": 388.0309,
      "step": 1020
    },
    {
      "epoch": 0.008322627041265686,
      "grad_norm": 11426.04296875,
      "learning_rate": 4.161616161616161e-06,
      "loss": 342.7975,
      "step": 1030
    },
    {
      "epoch": 0.008403429245549819,
      "grad_norm": 1648.6806640625,
      "learning_rate": 4.2020202020202026e-06,
      "loss": 377.397,
      "step": 1040
    },
    {
      "epoch": 0.008484231449833951,
      "grad_norm": 929.9481811523438,
      "learning_rate": 4.242424242424243e-06,
      "loss": 316.3978,
      "step": 1050
    },
    {
      "epoch": 0.008565033654118084,
      "grad_norm": 1129.7996826171875,
      "learning_rate": 4.282828282828283e-06,
      "loss": 299.9879,
      "step": 1060
    },
    {
      "epoch": 0.008645835858402217,
      "grad_norm": 1400.36376953125,
      "learning_rate": 4.323232323232323e-06,
      "loss": 350.1171,
      "step": 1070
    },
    {
      "epoch": 0.00872663806268635,
      "grad_norm": 1833.9061279296875,
      "learning_rate": 4.363636363636364e-06,
      "loss": 325.1123,
      "step": 1080
    },
    {
      "epoch": 0.008807440266970484,
      "grad_norm": 1330.033203125,
      "learning_rate": 4.4040404040404044e-06,
      "loss": 325.3823,
      "step": 1090
    },
    {
      "epoch": 0.008888242471254616,
      "grad_norm": 910.3088989257812,
      "learning_rate": 4.444444444444445e-06,
      "loss": 301.4958,
      "step": 1100
    },
    {
      "epoch": 0.008969044675538749,
      "grad_norm": 1182.816162109375,
      "learning_rate": 4.484848484848485e-06,
      "loss": 277.6404,
      "step": 1110
    },
    {
      "epoch": 0.009049846879822881,
      "grad_norm": 4707.03857421875,
      "learning_rate": 4.525252525252525e-06,
      "loss": 424.5438,
      "step": 1120
    },
    {
      "epoch": 0.009130649084107014,
      "grad_norm": 2059.0185546875,
      "learning_rate": 4.565656565656566e-06,
      "loss": 324.0304,
      "step": 1130
    },
    {
      "epoch": 0.009211451288391147,
      "grad_norm": 3679.1044921875,
      "learning_rate": 4.606060606060606e-06,
      "loss": 303.365,
      "step": 1140
    },
    {
      "epoch": 0.00929225349267528,
      "grad_norm": 2518.44970703125,
      "learning_rate": 4.646464646464647e-06,
      "loss": 257.8196,
      "step": 1150
    },
    {
      "epoch": 0.009373055696959414,
      "grad_norm": 1017.4381713867188,
      "learning_rate": 4.6868686868686874e-06,
      "loss": 308.7409,
      "step": 1160
    },
    {
      "epoch": 0.009453857901243546,
      "grad_norm": 1016.4625244140625,
      "learning_rate": 4.727272727272727e-06,
      "loss": 309.7148,
      "step": 1170
    },
    {
      "epoch": 0.009534660105527679,
      "grad_norm": 2000.1339111328125,
      "learning_rate": 4.767676767676768e-06,
      "loss": 416.5233,
      "step": 1180
    },
    {
      "epoch": 0.009615462309811812,
      "grad_norm": 1349.24755859375,
      "learning_rate": 4.808080808080808e-06,
      "loss": 312.0022,
      "step": 1190
    },
    {
      "epoch": 0.009696264514095944,
      "grad_norm": 605.5498046875,
      "learning_rate": 4.848484848484849e-06,
      "loss": 373.616,
      "step": 1200
    },
    {
      "epoch": 0.009777066718380077,
      "grad_norm": 1045.7061767578125,
      "learning_rate": 4.888888888888889e-06,
      "loss": 263.9121,
      "step": 1210
    },
    {
      "epoch": 0.00985786892266421,
      "grad_norm": 997.4849243164062,
      "learning_rate": 4.929292929292929e-06,
      "loss": 299.2633,
      "step": 1220
    },
    {
      "epoch": 0.009938671126948344,
      "grad_norm": 690.6622314453125,
      "learning_rate": 4.96969696969697e-06,
      "loss": 216.8853,
      "step": 1230
    },
    {
      "epoch": 0.010019473331232476,
      "grad_norm": 1591.2972412109375,
      "learning_rate": 5.010101010101011e-06,
      "loss": 339.3444,
      "step": 1240
    },
    {
      "epoch": 0.010100275535516609,
      "grad_norm": 749.3450317382812,
      "learning_rate": 5.050505050505051e-06,
      "loss": 315.4824,
      "step": 1250
    },
    {
      "epoch": 0.010181077739800742,
      "grad_norm": 923.2822265625,
      "learning_rate": 5.090909090909091e-06,
      "loss": 265.8641,
      "step": 1260
    },
    {
      "epoch": 0.010261879944084874,
      "grad_norm": 1161.5048828125,
      "learning_rate": 5.131313131313131e-06,
      "loss": 255.6592,
      "step": 1270
    },
    {
      "epoch": 0.010342682148369007,
      "grad_norm": 949.8252563476562,
      "learning_rate": 5.171717171717172e-06,
      "loss": 350.1398,
      "step": 1280
    },
    {
      "epoch": 0.01042348435265314,
      "grad_norm": 1018.5875854492188,
      "learning_rate": 5.212121212121213e-06,
      "loss": 313.8918,
      "step": 1290
    },
    {
      "epoch": 0.010504286556937274,
      "grad_norm": 3170.640869140625,
      "learning_rate": 5.2525252525252526e-06,
      "loss": 305.4674,
      "step": 1300
    },
    {
      "epoch": 0.010585088761221407,
      "grad_norm": 1600.813720703125,
      "learning_rate": 5.292929292929293e-06,
      "loss": 331.4024,
      "step": 1310
    },
    {
      "epoch": 0.01066589096550554,
      "grad_norm": 1063.93408203125,
      "learning_rate": 5.333333333333334e-06,
      "loss": 373.562,
      "step": 1320
    },
    {
      "epoch": 0.010746693169789672,
      "grad_norm": 665.6146240234375,
      "learning_rate": 5.373737373737374e-06,
      "loss": 235.0867,
      "step": 1330
    },
    {
      "epoch": 0.010827495374073804,
      "grad_norm": 1380.6151123046875,
      "learning_rate": 5.414141414141415e-06,
      "loss": 324.2148,
      "step": 1340
    },
    {
      "epoch": 0.010908297578357937,
      "grad_norm": 3627.843994140625,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 364.2515,
      "step": 1350
    },
    {
      "epoch": 0.01098909978264207,
      "grad_norm": 850.1503295898438,
      "learning_rate": 5.494949494949495e-06,
      "loss": 408.481,
      "step": 1360
    },
    {
      "epoch": 0.011069901986926204,
      "grad_norm": 2091.482421875,
      "learning_rate": 5.5353535353535355e-06,
      "loss": 267.4172,
      "step": 1370
    },
    {
      "epoch": 0.011150704191210337,
      "grad_norm": 1496.604248046875,
      "learning_rate": 5.575757575757576e-06,
      "loss": 295.7435,
      "step": 1380
    },
    {
      "epoch": 0.01123150639549447,
      "grad_norm": 2673.033203125,
      "learning_rate": 5.616161616161617e-06,
      "loss": 251.5225,
      "step": 1390
    },
    {
      "epoch": 0.011312308599778602,
      "grad_norm": 1121.48779296875,
      "learning_rate": 5.656565656565657e-06,
      "loss": 318.3079,
      "step": 1400
    },
    {
      "epoch": 0.011393110804062734,
      "grad_norm": 866.069091796875,
      "learning_rate": 5.696969696969697e-06,
      "loss": 283.861,
      "step": 1410
    },
    {
      "epoch": 0.011473913008346867,
      "grad_norm": 1132.9764404296875,
      "learning_rate": 5.7373737373737374e-06,
      "loss": 312.9168,
      "step": 1420
    },
    {
      "epoch": 0.011554715212631,
      "grad_norm": 1029.2037353515625,
      "learning_rate": 5.777777777777778e-06,
      "loss": 307.1583,
      "step": 1430
    },
    {
      "epoch": 0.011635517416915134,
      "grad_norm": 1138.4461669921875,
      "learning_rate": 5.8181818181818185e-06,
      "loss": 266.8528,
      "step": 1440
    },
    {
      "epoch": 0.011716319621199267,
      "grad_norm": 1278.30224609375,
      "learning_rate": 5.858585858585859e-06,
      "loss": 339.6735,
      "step": 1450
    },
    {
      "epoch": 0.0117971218254834,
      "grad_norm": 908.4046630859375,
      "learning_rate": 5.898989898989899e-06,
      "loss": 331.0705,
      "step": 1460
    },
    {
      "epoch": 0.011877924029767532,
      "grad_norm": 1165.220947265625,
      "learning_rate": 5.93939393939394e-06,
      "loss": 267.2519,
      "step": 1470
    },
    {
      "epoch": 0.011958726234051665,
      "grad_norm": 3219.732177734375,
      "learning_rate": 5.979797979797981e-06,
      "loss": 485.5872,
      "step": 1480
    },
    {
      "epoch": 0.012039528438335797,
      "grad_norm": 1050.6778564453125,
      "learning_rate": 6.0202020202020204e-06,
      "loss": 325.8213,
      "step": 1490
    },
    {
      "epoch": 0.01212033064261993,
      "grad_norm": 1031.156005859375,
      "learning_rate": 6.060606060606061e-06,
      "loss": 302.1713,
      "step": 1500
    },
    {
      "epoch": 0.012201132846904064,
      "grad_norm": 1457.485107421875,
      "learning_rate": 6.101010101010101e-06,
      "loss": 267.4604,
      "step": 1510
    },
    {
      "epoch": 0.012281935051188197,
      "grad_norm": 1509.94091796875,
      "learning_rate": 6.141414141414142e-06,
      "loss": 368.4667,
      "step": 1520
    },
    {
      "epoch": 0.01236273725547233,
      "grad_norm": 1406.15673828125,
      "learning_rate": 6.181818181818183e-06,
      "loss": 349.9011,
      "step": 1530
    },
    {
      "epoch": 0.012443539459756462,
      "grad_norm": 1459.2613525390625,
      "learning_rate": 6.222222222222222e-06,
      "loss": 325.6512,
      "step": 1540
    },
    {
      "epoch": 0.012524341664040595,
      "grad_norm": 3242.78271484375,
      "learning_rate": 6.262626262626263e-06,
      "loss": 375.0032,
      "step": 1550
    },
    {
      "epoch": 0.012605143868324727,
      "grad_norm": 2443.515625,
      "learning_rate": 6.303030303030303e-06,
      "loss": 379.789,
      "step": 1560
    },
    {
      "epoch": 0.01268594607260886,
      "grad_norm": 1770.516845703125,
      "learning_rate": 6.343434343434344e-06,
      "loss": 251.652,
      "step": 1570
    },
    {
      "epoch": 0.012766748276892994,
      "grad_norm": 1482.888671875,
      "learning_rate": 6.383838383838384e-06,
      "loss": 324.2996,
      "step": 1580
    },
    {
      "epoch": 0.012847550481177127,
      "grad_norm": 1663.3935546875,
      "learning_rate": 6.424242424242424e-06,
      "loss": 301.8472,
      "step": 1590
    },
    {
      "epoch": 0.01292835268546126,
      "grad_norm": 23524.42578125,
      "learning_rate": 6.464646464646465e-06,
      "loss": 335.0892,
      "step": 1600
    },
    {
      "epoch": 0.013009154889745392,
      "grad_norm": 862.7949829101562,
      "learning_rate": 6.505050505050505e-06,
      "loss": 306.2031,
      "step": 1610
    },
    {
      "epoch": 0.013089957094029525,
      "grad_norm": 2391.976806640625,
      "learning_rate": 6.545454545454547e-06,
      "loss": 310.1344,
      "step": 1620
    },
    {
      "epoch": 0.013170759298313657,
      "grad_norm": 1260.0029296875,
      "learning_rate": 6.5858585858585856e-06,
      "loss": 372.1161,
      "step": 1630
    },
    {
      "epoch": 0.01325156150259779,
      "grad_norm": 1587.1514892578125,
      "learning_rate": 6.626262626262626e-06,
      "loss": 351.7325,
      "step": 1640
    },
    {
      "epoch": 0.013332363706881924,
      "grad_norm": 1152.1556396484375,
      "learning_rate": 6.666666666666667e-06,
      "loss": 373.3706,
      "step": 1650
    },
    {
      "epoch": 0.013413165911166057,
      "grad_norm": 1005.15771484375,
      "learning_rate": 6.707070707070707e-06,
      "loss": 317.8097,
      "step": 1660
    },
    {
      "epoch": 0.01349396811545019,
      "grad_norm": 1090.3779296875,
      "learning_rate": 6.747474747474749e-06,
      "loss": 315.5722,
      "step": 1670
    },
    {
      "epoch": 0.013574770319734322,
      "grad_norm": 1011.2723388671875,
      "learning_rate": 6.787878787878789e-06,
      "loss": 231.6832,
      "step": 1680
    },
    {
      "epoch": 0.013655572524018455,
      "grad_norm": 798.0405883789062,
      "learning_rate": 6.828282828282828e-06,
      "loss": 306.8582,
      "step": 1690
    },
    {
      "epoch": 0.013736374728302587,
      "grad_norm": 855.2308959960938,
      "learning_rate": 6.8686868686868685e-06,
      "loss": 282.2944,
      "step": 1700
    },
    {
      "epoch": 0.01381717693258672,
      "grad_norm": 1325.1092529296875,
      "learning_rate": 6.909090909090909e-06,
      "loss": 427.1749,
      "step": 1710
    },
    {
      "epoch": 0.013897979136870854,
      "grad_norm": 1027.2860107421875,
      "learning_rate": 6.9494949494949505e-06,
      "loss": 232.57,
      "step": 1720
    },
    {
      "epoch": 0.013978781341154987,
      "grad_norm": 1047.0118408203125,
      "learning_rate": 6.989898989898991e-06,
      "loss": 272.4593,
      "step": 1730
    },
    {
      "epoch": 0.01405958354543912,
      "grad_norm": 1453.4931640625,
      "learning_rate": 7.03030303030303e-06,
      "loss": 287.978,
      "step": 1740
    },
    {
      "epoch": 0.014140385749723252,
      "grad_norm": 1631.84521484375,
      "learning_rate": 7.0707070707070704e-06,
      "loss": 222.9585,
      "step": 1750
    },
    {
      "epoch": 0.014221187954007385,
      "grad_norm": 1109.9012451171875,
      "learning_rate": 7.111111111111112e-06,
      "loss": 263.4153,
      "step": 1760
    },
    {
      "epoch": 0.014301990158291518,
      "grad_norm": 1374.5731201171875,
      "learning_rate": 7.151515151515152e-06,
      "loss": 224.7149,
      "step": 1770
    },
    {
      "epoch": 0.01438279236257565,
      "grad_norm": 700.2552490234375,
      "learning_rate": 7.191919191919193e-06,
      "loss": 243.2667,
      "step": 1780
    },
    {
      "epoch": 0.014463594566859785,
      "grad_norm": 689.7608032226562,
      "learning_rate": 7.232323232323232e-06,
      "loss": 209.3831,
      "step": 1790
    },
    {
      "epoch": 0.014544396771143917,
      "grad_norm": 1072.593994140625,
      "learning_rate": 7.272727272727272e-06,
      "loss": 233.4359,
      "step": 1800
    },
    {
      "epoch": 0.01462519897542805,
      "grad_norm": 783.2555541992188,
      "learning_rate": 7.313131313131314e-06,
      "loss": 259.1145,
      "step": 1810
    },
    {
      "epoch": 0.014706001179712182,
      "grad_norm": 1364.049560546875,
      "learning_rate": 7.353535353535354e-06,
      "loss": 244.7944,
      "step": 1820
    },
    {
      "epoch": 0.014786803383996315,
      "grad_norm": 1947.3690185546875,
      "learning_rate": 7.393939393939395e-06,
      "loss": 276.0544,
      "step": 1830
    },
    {
      "epoch": 0.014867605588280448,
      "grad_norm": 1572.007568359375,
      "learning_rate": 7.434343434343435e-06,
      "loss": 310.638,
      "step": 1840
    },
    {
      "epoch": 0.01494840779256458,
      "grad_norm": 1419.4361572265625,
      "learning_rate": 7.474747474747475e-06,
      "loss": 409.634,
      "step": 1850
    },
    {
      "epoch": 0.015029209996848715,
      "grad_norm": 1355.0137939453125,
      "learning_rate": 7.515151515151516e-06,
      "loss": 253.9533,
      "step": 1860
    },
    {
      "epoch": 0.015110012201132847,
      "grad_norm": 768.96923828125,
      "learning_rate": 7.555555555555556e-06,
      "loss": 304.4757,
      "step": 1870
    },
    {
      "epoch": 0.01519081440541698,
      "grad_norm": 959.4989624023438,
      "learning_rate": 7.595959595959597e-06,
      "loss": 345.032,
      "step": 1880
    },
    {
      "epoch": 0.015271616609701113,
      "grad_norm": 1339.84228515625,
      "learning_rate": 7.636363636363638e-06,
      "loss": 296.0798,
      "step": 1890
    },
    {
      "epoch": 0.015352418813985245,
      "grad_norm": 1036.2491455078125,
      "learning_rate": 7.676767676767677e-06,
      "loss": 300.197,
      "step": 1900
    },
    {
      "epoch": 0.015433221018269378,
      "grad_norm": 707.27587890625,
      "learning_rate": 7.717171717171717e-06,
      "loss": 277.1603,
      "step": 1910
    },
    {
      "epoch": 0.015514023222553512,
      "grad_norm": 1203.453857421875,
      "learning_rate": 7.757575757575758e-06,
      "loss": 303.6785,
      "step": 1920
    },
    {
      "epoch": 0.015594825426837645,
      "grad_norm": 1172.7025146484375,
      "learning_rate": 7.797979797979799e-06,
      "loss": 246.9346,
      "step": 1930
    },
    {
      "epoch": 0.015675627631121777,
      "grad_norm": 1082.4605712890625,
      "learning_rate": 7.838383838383839e-06,
      "loss": 296.1259,
      "step": 1940
    },
    {
      "epoch": 0.01575642983540591,
      "grad_norm": 1456.42529296875,
      "learning_rate": 7.878787878787878e-06,
      "loss": 269.7506,
      "step": 1950
    },
    {
      "epoch": 0.015837232039690043,
      "grad_norm": 1794.72119140625,
      "learning_rate": 7.919191919191919e-06,
      "loss": 246.411,
      "step": 1960
    },
    {
      "epoch": 0.015918034243974175,
      "grad_norm": 3157.114990234375,
      "learning_rate": 7.959595959595959e-06,
      "loss": 256.127,
      "step": 1970
    },
    {
      "epoch": 0.015998836448258308,
      "grad_norm": 1361.6929931640625,
      "learning_rate": 8.000000000000001e-06,
      "loss": 308.1984,
      "step": 1980
    },
    {
      "epoch": 0.01607963865254244,
      "grad_norm": 1006.965087890625,
      "learning_rate": 8.040404040404042e-06,
      "loss": 341.8693,
      "step": 1990
    },
    {
      "epoch": 0.016160440856826573,
      "grad_norm": 1541.38720703125,
      "learning_rate": 8.080808080808082e-06,
      "loss": 274.4633,
      "step": 2000
    },
    {
      "epoch": 0.016241243061110706,
      "grad_norm": 1735.104248046875,
      "learning_rate": 8.121212121212121e-06,
      "loss": 258.4104,
      "step": 2010
    },
    {
      "epoch": 0.01632204526539484,
      "grad_norm": 2176.154052734375,
      "learning_rate": 8.161616161616162e-06,
      "loss": 314.5508,
      "step": 2020
    },
    {
      "epoch": 0.016402847469678974,
      "grad_norm": 1052.533447265625,
      "learning_rate": 8.202020202020202e-06,
      "loss": 312.927,
      "step": 2030
    },
    {
      "epoch": 0.016483649673963107,
      "grad_norm": 1208.69189453125,
      "learning_rate": 8.242424242424243e-06,
      "loss": 254.359,
      "step": 2040
    },
    {
      "epoch": 0.01656445187824724,
      "grad_norm": 1978.5025634765625,
      "learning_rate": 8.282828282828283e-06,
      "loss": 299.8744,
      "step": 2050
    },
    {
      "epoch": 0.016645254082531372,
      "grad_norm": 3605.6494140625,
      "learning_rate": 8.323232323232322e-06,
      "loss": 298.715,
      "step": 2060
    },
    {
      "epoch": 0.016726056286815505,
      "grad_norm": 1599.973876953125,
      "learning_rate": 8.363636363636365e-06,
      "loss": 258.9758,
      "step": 2070
    },
    {
      "epoch": 0.016806858491099638,
      "grad_norm": 1183.451904296875,
      "learning_rate": 8.404040404040405e-06,
      "loss": 352.8176,
      "step": 2080
    },
    {
      "epoch": 0.01688766069538377,
      "grad_norm": 1582.7120361328125,
      "learning_rate": 8.444444444444446e-06,
      "loss": 309.3484,
      "step": 2090
    },
    {
      "epoch": 0.016968462899667903,
      "grad_norm": 932.9716796875,
      "learning_rate": 8.484848484848486e-06,
      "loss": 255.265,
      "step": 2100
    },
    {
      "epoch": 0.017049265103952035,
      "grad_norm": 922.5059814453125,
      "learning_rate": 8.525252525252525e-06,
      "loss": 253.6583,
      "step": 2110
    },
    {
      "epoch": 0.017130067308236168,
      "grad_norm": 1196.361083984375,
      "learning_rate": 8.565656565656566e-06,
      "loss": 323.4844,
      "step": 2120
    },
    {
      "epoch": 0.0172108695125203,
      "grad_norm": 1005.9546508789062,
      "learning_rate": 8.606060606060606e-06,
      "loss": 351.4896,
      "step": 2130
    },
    {
      "epoch": 0.017291671716804433,
      "grad_norm": 1585.8636474609375,
      "learning_rate": 8.646464646464647e-06,
      "loss": 325.0701,
      "step": 2140
    },
    {
      "epoch": 0.017372473921088566,
      "grad_norm": 3758.6982421875,
      "learning_rate": 8.686868686868687e-06,
      "loss": 203.9509,
      "step": 2150
    },
    {
      "epoch": 0.0174532761253727,
      "grad_norm": 1602.480224609375,
      "learning_rate": 8.727272727272728e-06,
      "loss": 294.8089,
      "step": 2160
    },
    {
      "epoch": 0.017534078329656835,
      "grad_norm": 1571.1812744140625,
      "learning_rate": 8.767676767676768e-06,
      "loss": 220.656,
      "step": 2170
    },
    {
      "epoch": 0.017614880533940967,
      "grad_norm": 1073.2261962890625,
      "learning_rate": 8.808080808080809e-06,
      "loss": 221.6837,
      "step": 2180
    },
    {
      "epoch": 0.0176956827382251,
      "grad_norm": 1125.1983642578125,
      "learning_rate": 8.84848484848485e-06,
      "loss": 286.0061,
      "step": 2190
    },
    {
      "epoch": 0.017776484942509233,
      "grad_norm": 1008.795654296875,
      "learning_rate": 8.88888888888889e-06,
      "loss": 282.5281,
      "step": 2200
    },
    {
      "epoch": 0.017857287146793365,
      "grad_norm": 1810.7894287109375,
      "learning_rate": 8.92929292929293e-06,
      "loss": 226.4816,
      "step": 2210
    },
    {
      "epoch": 0.017938089351077498,
      "grad_norm": 894.2589721679688,
      "learning_rate": 8.96969696969697e-06,
      "loss": 251.6401,
      "step": 2220
    },
    {
      "epoch": 0.01801889155536163,
      "grad_norm": 1232.9827880859375,
      "learning_rate": 9.01010101010101e-06,
      "loss": 248.7544,
      "step": 2230
    },
    {
      "epoch": 0.018099693759645763,
      "grad_norm": 1993.9267578125,
      "learning_rate": 9.05050505050505e-06,
      "loss": 256.8296,
      "step": 2240
    },
    {
      "epoch": 0.018180495963929896,
      "grad_norm": 967.433837890625,
      "learning_rate": 9.090909090909091e-06,
      "loss": 245.6321,
      "step": 2250
    },
    {
      "epoch": 0.018261298168214028,
      "grad_norm": 2560.1728515625,
      "learning_rate": 9.131313131313132e-06,
      "loss": 234.9478,
      "step": 2260
    },
    {
      "epoch": 0.01834210037249816,
      "grad_norm": 590.1747436523438,
      "learning_rate": 9.171717171717172e-06,
      "loss": 236.6796,
      "step": 2270
    },
    {
      "epoch": 0.018422902576782293,
      "grad_norm": 1504.942626953125,
      "learning_rate": 9.212121212121213e-06,
      "loss": 367.3362,
      "step": 2280
    },
    {
      "epoch": 0.018503704781066426,
      "grad_norm": 977.7069091796875,
      "learning_rate": 9.252525252525253e-06,
      "loss": 276.6067,
      "step": 2290
    },
    {
      "epoch": 0.01858450698535056,
      "grad_norm": 1384.83203125,
      "learning_rate": 9.292929292929294e-06,
      "loss": 340.7896,
      "step": 2300
    },
    {
      "epoch": 0.018665309189634695,
      "grad_norm": 1302.6343994140625,
      "learning_rate": 9.333333333333334e-06,
      "loss": 198.1527,
      "step": 2310
    },
    {
      "epoch": 0.018746111393918827,
      "grad_norm": 836.4732666015625,
      "learning_rate": 9.373737373737375e-06,
      "loss": 209.2565,
      "step": 2320
    },
    {
      "epoch": 0.01882691359820296,
      "grad_norm": 1400.0604248046875,
      "learning_rate": 9.414141414141414e-06,
      "loss": 305.2968,
      "step": 2330
    },
    {
      "epoch": 0.018907715802487093,
      "grad_norm": 1454.78125,
      "learning_rate": 9.454545454545454e-06,
      "loss": 260.4474,
      "step": 2340
    },
    {
      "epoch": 0.018988518006771225,
      "grad_norm": 4679.6923828125,
      "learning_rate": 9.494949494949495e-06,
      "loss": 247.9314,
      "step": 2350
    },
    {
      "epoch": 0.019069320211055358,
      "grad_norm": 6640.1201171875,
      "learning_rate": 9.535353535353535e-06,
      "loss": 259.6065,
      "step": 2360
    },
    {
      "epoch": 0.01915012241533949,
      "grad_norm": 1564.781982421875,
      "learning_rate": 9.575757575757578e-06,
      "loss": 245.3016,
      "step": 2370
    },
    {
      "epoch": 0.019230924619623623,
      "grad_norm": 2078.74267578125,
      "learning_rate": 9.616161616161616e-06,
      "loss": 284.7133,
      "step": 2380
    },
    {
      "epoch": 0.019311726823907756,
      "grad_norm": 1441.0360107421875,
      "learning_rate": 9.656565656565657e-06,
      "loss": 270.4787,
      "step": 2390
    },
    {
      "epoch": 0.01939252902819189,
      "grad_norm": 744.0514526367188,
      "learning_rate": 9.696969696969698e-06,
      "loss": 200.2359,
      "step": 2400
    },
    {
      "epoch": 0.01947333123247602,
      "grad_norm": 1168.913818359375,
      "learning_rate": 9.737373737373738e-06,
      "loss": 269.4149,
      "step": 2410
    },
    {
      "epoch": 0.019554133436760154,
      "grad_norm": 3848.146484375,
      "learning_rate": 9.777777777777779e-06,
      "loss": 250.6479,
      "step": 2420
    },
    {
      "epoch": 0.019634935641044286,
      "grad_norm": 798.60595703125,
      "learning_rate": 9.818181818181818e-06,
      "loss": 245.7341,
      "step": 2430
    },
    {
      "epoch": 0.01971573784532842,
      "grad_norm": 2139.72265625,
      "learning_rate": 9.858585858585858e-06,
      "loss": 275.653,
      "step": 2440
    },
    {
      "epoch": 0.019796540049612555,
      "grad_norm": 1223.6392822265625,
      "learning_rate": 9.898989898989899e-06,
      "loss": 271.2685,
      "step": 2450
    },
    {
      "epoch": 0.019877342253896688,
      "grad_norm": 1007.25439453125,
      "learning_rate": 9.93939393939394e-06,
      "loss": 235.147,
      "step": 2460
    },
    {
      "epoch": 0.01995814445818082,
      "grad_norm": 1195.012939453125,
      "learning_rate": 9.979797979797981e-06,
      "loss": 302.4475,
      "step": 2470
    },
    {
      "epoch": 0.020038946662464953,
      "grad_norm": 1530.9473876953125,
      "learning_rate": 1.0020202020202022e-05,
      "loss": 259.1597,
      "step": 2480
    },
    {
      "epoch": 0.020119748866749085,
      "grad_norm": 1403.546142578125,
      "learning_rate": 1.006060606060606e-05,
      "loss": 348.2433,
      "step": 2490
    },
    {
      "epoch": 0.020200551071033218,
      "grad_norm": 1328.4873046875,
      "learning_rate": 1.0101010101010101e-05,
      "loss": 289.0594,
      "step": 2500
    },
    {
      "epoch": 0.02028135327531735,
      "grad_norm": 1171.5048828125,
      "learning_rate": 1.0141414141414142e-05,
      "loss": 197.3946,
      "step": 2510
    },
    {
      "epoch": 0.020362155479601483,
      "grad_norm": 1274.6544189453125,
      "learning_rate": 1.0181818181818182e-05,
      "loss": 295.2945,
      "step": 2520
    },
    {
      "epoch": 0.020442957683885616,
      "grad_norm": 2023.71337890625,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 271.9707,
      "step": 2530
    },
    {
      "epoch": 0.02052375988816975,
      "grad_norm": 1765.538818359375,
      "learning_rate": 1.0262626262626262e-05,
      "loss": 246.6141,
      "step": 2540
    },
    {
      "epoch": 0.02060456209245388,
      "grad_norm": 859.3914794921875,
      "learning_rate": 1.0303030303030304e-05,
      "loss": 276.1375,
      "step": 2550
    },
    {
      "epoch": 0.020685364296738014,
      "grad_norm": 1024.8955078125,
      "learning_rate": 1.0343434343434345e-05,
      "loss": 239.6529,
      "step": 2560
    },
    {
      "epoch": 0.020766166501022146,
      "grad_norm": 1012.91455078125,
      "learning_rate": 1.0383838383838385e-05,
      "loss": 199.6656,
      "step": 2570
    },
    {
      "epoch": 0.02084696870530628,
      "grad_norm": 1371.8551025390625,
      "learning_rate": 1.0424242424242426e-05,
      "loss": 247.0821,
      "step": 2580
    },
    {
      "epoch": 0.020927770909590415,
      "grad_norm": 1338.343017578125,
      "learning_rate": 1.0464646464646465e-05,
      "loss": 260.7634,
      "step": 2590
    },
    {
      "epoch": 0.021008573113874548,
      "grad_norm": 917.8023071289062,
      "learning_rate": 1.0505050505050505e-05,
      "loss": 255.2869,
      "step": 2600
    },
    {
      "epoch": 0.02108937531815868,
      "grad_norm": 1167.3427734375,
      "learning_rate": 1.0545454545454546e-05,
      "loss": 255.108,
      "step": 2610
    },
    {
      "epoch": 0.021170177522442813,
      "grad_norm": 1662.1556396484375,
      "learning_rate": 1.0585858585858586e-05,
      "loss": 304.7596,
      "step": 2620
    },
    {
      "epoch": 0.021250979726726946,
      "grad_norm": 1393.7713623046875,
      "learning_rate": 1.0626262626262627e-05,
      "loss": 298.9948,
      "step": 2630
    },
    {
      "epoch": 0.02133178193101108,
      "grad_norm": 4169.89306640625,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 256.329,
      "step": 2640
    },
    {
      "epoch": 0.02141258413529521,
      "grad_norm": 961.8526000976562,
      "learning_rate": 1.0707070707070708e-05,
      "loss": 268.2124,
      "step": 2650
    },
    {
      "epoch": 0.021493386339579344,
      "grad_norm": 2141.90869140625,
      "learning_rate": 1.0747474747474748e-05,
      "loss": 279.6773,
      "step": 2660
    },
    {
      "epoch": 0.021574188543863476,
      "grad_norm": 1454.77392578125,
      "learning_rate": 1.0787878787878789e-05,
      "loss": 250.6374,
      "step": 2670
    },
    {
      "epoch": 0.02165499074814761,
      "grad_norm": 1119.782958984375,
      "learning_rate": 1.082828282828283e-05,
      "loss": 299.24,
      "step": 2680
    },
    {
      "epoch": 0.02173579295243174,
      "grad_norm": 1507.88916015625,
      "learning_rate": 1.086868686868687e-05,
      "loss": 304.945,
      "step": 2690
    },
    {
      "epoch": 0.021816595156715874,
      "grad_norm": 1235.4326171875,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 329.6972,
      "step": 2700
    },
    {
      "epoch": 0.021897397361000007,
      "grad_norm": 1516.6436767578125,
      "learning_rate": 1.094949494949495e-05,
      "loss": 276.193,
      "step": 2710
    },
    {
      "epoch": 0.02197819956528414,
      "grad_norm": 1332.3309326171875,
      "learning_rate": 1.098989898989899e-05,
      "loss": 309.7016,
      "step": 2720
    },
    {
      "epoch": 0.022059001769568275,
      "grad_norm": 1349.0360107421875,
      "learning_rate": 1.103030303030303e-05,
      "loss": 423.0254,
      "step": 2730
    },
    {
      "epoch": 0.022139803973852408,
      "grad_norm": 2262.348876953125,
      "learning_rate": 1.1070707070707071e-05,
      "loss": 260.9447,
      "step": 2740
    },
    {
      "epoch": 0.02222060617813654,
      "grad_norm": 1374.3009033203125,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 283.2908,
      "step": 2750
    },
    {
      "epoch": 0.022301408382420673,
      "grad_norm": 768.2625732421875,
      "learning_rate": 1.1151515151515152e-05,
      "loss": 243.0171,
      "step": 2760
    },
    {
      "epoch": 0.022382210586704806,
      "grad_norm": 4175.06396484375,
      "learning_rate": 1.1191919191919193e-05,
      "loss": 236.5765,
      "step": 2770
    },
    {
      "epoch": 0.02246301279098894,
      "grad_norm": 1220.933837890625,
      "learning_rate": 1.1232323232323233e-05,
      "loss": 254.5181,
      "step": 2780
    },
    {
      "epoch": 0.02254381499527307,
      "grad_norm": 1589.7581787109375,
      "learning_rate": 1.1272727272727274e-05,
      "loss": 183.8809,
      "step": 2790
    },
    {
      "epoch": 0.022624617199557204,
      "grad_norm": 1031.3692626953125,
      "learning_rate": 1.1313131313131314e-05,
      "loss": 246.612,
      "step": 2800
    },
    {
      "epoch": 0.022705419403841336,
      "grad_norm": 1396.37744140625,
      "learning_rate": 1.1353535353535353e-05,
      "loss": 224.2803,
      "step": 2810
    },
    {
      "epoch": 0.02278622160812547,
      "grad_norm": 1879.5634765625,
      "learning_rate": 1.1393939393939394e-05,
      "loss": 287.104,
      "step": 2820
    },
    {
      "epoch": 0.0228670238124096,
      "grad_norm": 1376.0625,
      "learning_rate": 1.1434343434343434e-05,
      "loss": 284.0272,
      "step": 2830
    },
    {
      "epoch": 0.022947826016693734,
      "grad_norm": 682.961181640625,
      "learning_rate": 1.1474747474747475e-05,
      "loss": 210.3876,
      "step": 2840
    },
    {
      "epoch": 0.023028628220977867,
      "grad_norm": 1677.2169189453125,
      "learning_rate": 1.1515151515151517e-05,
      "loss": 326.8289,
      "step": 2850
    },
    {
      "epoch": 0.023109430425262,
      "grad_norm": 733.2987060546875,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 230.43,
      "step": 2860
    },
    {
      "epoch": 0.023190232629546136,
      "grad_norm": 630.212890625,
      "learning_rate": 1.1595959595959597e-05,
      "loss": 205.1732,
      "step": 2870
    },
    {
      "epoch": 0.023271034833830268,
      "grad_norm": 1535.36572265625,
      "learning_rate": 1.1636363636363637e-05,
      "loss": 304.9371,
      "step": 2880
    },
    {
      "epoch": 0.0233518370381144,
      "grad_norm": 1065.3255615234375,
      "learning_rate": 1.1676767676767678e-05,
      "loss": 219.3441,
      "step": 2890
    },
    {
      "epoch": 0.023432639242398533,
      "grad_norm": 2319.004638671875,
      "learning_rate": 1.1717171717171718e-05,
      "loss": 274.9734,
      "step": 2900
    },
    {
      "epoch": 0.023513441446682666,
      "grad_norm": 1362.914794921875,
      "learning_rate": 1.1757575757575757e-05,
      "loss": 214.0071,
      "step": 2910
    },
    {
      "epoch": 0.0235942436509668,
      "grad_norm": 1748.42333984375,
      "learning_rate": 1.1797979797979798e-05,
      "loss": 247.8715,
      "step": 2920
    },
    {
      "epoch": 0.02367504585525093,
      "grad_norm": 1357.4864501953125,
      "learning_rate": 1.1838383838383838e-05,
      "loss": 367.5896,
      "step": 2930
    },
    {
      "epoch": 0.023755848059535064,
      "grad_norm": 1028.1129150390625,
      "learning_rate": 1.187878787878788e-05,
      "loss": 248.4116,
      "step": 2940
    },
    {
      "epoch": 0.023836650263819197,
      "grad_norm": 1497.2218017578125,
      "learning_rate": 1.1919191919191921e-05,
      "loss": 309.2206,
      "step": 2950
    },
    {
      "epoch": 0.02391745246810333,
      "grad_norm": 830.5894775390625,
      "learning_rate": 1.1959595959595961e-05,
      "loss": 203.645,
      "step": 2960
    },
    {
      "epoch": 0.023998254672387462,
      "grad_norm": 1067.825439453125,
      "learning_rate": 1.2e-05,
      "loss": 226.1753,
      "step": 2970
    },
    {
      "epoch": 0.024079056876671594,
      "grad_norm": 1088.949462890625,
      "learning_rate": 1.2040404040404041e-05,
      "loss": 235.4068,
      "step": 2980
    },
    {
      "epoch": 0.024159859080955727,
      "grad_norm": 1378.7232666015625,
      "learning_rate": 1.2080808080808081e-05,
      "loss": 258.9663,
      "step": 2990
    },
    {
      "epoch": 0.02424066128523986,
      "grad_norm": 1665.1182861328125,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 178.7374,
      "step": 3000
    },
    {
      "epoch": 0.024321463489523996,
      "grad_norm": 1254.1650390625,
      "learning_rate": 1.2161616161616162e-05,
      "loss": 196.0946,
      "step": 3010
    },
    {
      "epoch": 0.02440226569380813,
      "grad_norm": 1680.0679931640625,
      "learning_rate": 1.2202020202020201e-05,
      "loss": 232.687,
      "step": 3020
    },
    {
      "epoch": 0.02448306789809226,
      "grad_norm": 1063.8585205078125,
      "learning_rate": 1.2242424242424242e-05,
      "loss": 247.0947,
      "step": 3030
    },
    {
      "epoch": 0.024563870102376394,
      "grad_norm": 3708.090087890625,
      "learning_rate": 1.2282828282828284e-05,
      "loss": 252.8591,
      "step": 3040
    },
    {
      "epoch": 0.024644672306660526,
      "grad_norm": 1465.7147216796875,
      "learning_rate": 1.2323232323232325e-05,
      "loss": 262.5838,
      "step": 3050
    },
    {
      "epoch": 0.02472547451094466,
      "grad_norm": 1327.937255859375,
      "learning_rate": 1.2363636363636365e-05,
      "loss": 363.5593,
      "step": 3060
    },
    {
      "epoch": 0.02480627671522879,
      "grad_norm": 1344.0130615234375,
      "learning_rate": 1.2404040404040404e-05,
      "loss": 257.7762,
      "step": 3070
    },
    {
      "epoch": 0.024887078919512924,
      "grad_norm": 1054.320068359375,
      "learning_rate": 1.2444444444444445e-05,
      "loss": 247.01,
      "step": 3080
    },
    {
      "epoch": 0.024967881123797057,
      "grad_norm": 1068.103515625,
      "learning_rate": 1.2484848484848485e-05,
      "loss": 248.3725,
      "step": 3090
    },
    {
      "epoch": 0.02504868332808119,
      "grad_norm": 1697.66259765625,
      "learning_rate": 1.2525252525252526e-05,
      "loss": 258.5855,
      "step": 3100
    },
    {
      "epoch": 0.025129485532365322,
      "grad_norm": 2154.507080078125,
      "learning_rate": 1.2565656565656566e-05,
      "loss": 324.4505,
      "step": 3110
    },
    {
      "epoch": 0.025210287736649455,
      "grad_norm": 1405.111083984375,
      "learning_rate": 1.2606060606060607e-05,
      "loss": 201.9885,
      "step": 3120
    },
    {
      "epoch": 0.025291089940933587,
      "grad_norm": 1943.2344970703125,
      "learning_rate": 1.2646464646464647e-05,
      "loss": 279.3126,
      "step": 3130
    },
    {
      "epoch": 0.02537189214521772,
      "grad_norm": 1883.3538818359375,
      "learning_rate": 1.2686868686868688e-05,
      "loss": 305.4725,
      "step": 3140
    },
    {
      "epoch": 0.025452694349501856,
      "grad_norm": 2943.758544921875,
      "learning_rate": 1.2727272727272727e-05,
      "loss": 282.1084,
      "step": 3150
    },
    {
      "epoch": 0.02553349655378599,
      "grad_norm": 1242.160400390625,
      "learning_rate": 1.2767676767676767e-05,
      "loss": 220.8162,
      "step": 3160
    },
    {
      "epoch": 0.02561429875807012,
      "grad_norm": 2627.211181640625,
      "learning_rate": 1.2808080808080808e-05,
      "loss": 303.4214,
      "step": 3170
    },
    {
      "epoch": 0.025695100962354254,
      "grad_norm": 1310.1988525390625,
      "learning_rate": 1.2848484848484848e-05,
      "loss": 259.6753,
      "step": 3180
    },
    {
      "epoch": 0.025775903166638386,
      "grad_norm": 1910.4666748046875,
      "learning_rate": 1.2888888888888889e-05,
      "loss": 273.7454,
      "step": 3190
    },
    {
      "epoch": 0.02585670537092252,
      "grad_norm": 740.5687255859375,
      "learning_rate": 1.292929292929293e-05,
      "loss": 292.6491,
      "step": 3200
    },
    {
      "epoch": 0.02593750757520665,
      "grad_norm": 1202.99462890625,
      "learning_rate": 1.296969696969697e-05,
      "loss": 243.2686,
      "step": 3210
    },
    {
      "epoch": 0.026018309779490784,
      "grad_norm": 2174.3525390625,
      "learning_rate": 1.301010101010101e-05,
      "loss": 276.3461,
      "step": 3220
    },
    {
      "epoch": 0.026099111983774917,
      "grad_norm": 1177.9141845703125,
      "learning_rate": 1.3050505050505051e-05,
      "loss": 246.7616,
      "step": 3230
    },
    {
      "epoch": 0.02617991418805905,
      "grad_norm": 1337.3155517578125,
      "learning_rate": 1.3090909090909093e-05,
      "loss": 279.5415,
      "step": 3240
    },
    {
      "epoch": 0.026260716392343182,
      "grad_norm": 970.5191040039062,
      "learning_rate": 1.3131313131313134e-05,
      "loss": 240.0207,
      "step": 3250
    },
    {
      "epoch": 0.026341518596627315,
      "grad_norm": 1119.1689453125,
      "learning_rate": 1.3171717171717171e-05,
      "loss": 242.7696,
      "step": 3260
    },
    {
      "epoch": 0.026422320800911447,
      "grad_norm": 4005.26318359375,
      "learning_rate": 1.3212121212121212e-05,
      "loss": 288.4158,
      "step": 3270
    },
    {
      "epoch": 0.02650312300519558,
      "grad_norm": 2148.187255859375,
      "learning_rate": 1.3252525252525252e-05,
      "loss": 224.5919,
      "step": 3280
    },
    {
      "epoch": 0.026583925209479716,
      "grad_norm": 1367.2222900390625,
      "learning_rate": 1.3292929292929293e-05,
      "loss": 212.8641,
      "step": 3290
    },
    {
      "epoch": 0.02666472741376385,
      "grad_norm": 1278.0506591796875,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 222.3518,
      "step": 3300
    },
    {
      "epoch": 0.02674552961804798,
      "grad_norm": 1361.965087890625,
      "learning_rate": 1.3373737373737374e-05,
      "loss": 257.0285,
      "step": 3310
    },
    {
      "epoch": 0.026826331822332114,
      "grad_norm": 1330.8619384765625,
      "learning_rate": 1.3414141414141414e-05,
      "loss": 293.7213,
      "step": 3320
    },
    {
      "epoch": 0.026907134026616247,
      "grad_norm": 1107.8526611328125,
      "learning_rate": 1.3454545454545457e-05,
      "loss": 198.6048,
      "step": 3330
    },
    {
      "epoch": 0.02698793623090038,
      "grad_norm": 1048.5009765625,
      "learning_rate": 1.3494949494949497e-05,
      "loss": 197.4301,
      "step": 3340
    },
    {
      "epoch": 0.027068738435184512,
      "grad_norm": 2446.47119140625,
      "learning_rate": 1.3535353535353538e-05,
      "loss": 287.2842,
      "step": 3350
    },
    {
      "epoch": 0.027149540639468644,
      "grad_norm": 1396.544921875,
      "learning_rate": 1.3575757575757578e-05,
      "loss": 258.2231,
      "step": 3360
    },
    {
      "epoch": 0.027230342843752777,
      "grad_norm": 904.5388793945312,
      "learning_rate": 1.3616161616161615e-05,
      "loss": 376.3808,
      "step": 3370
    },
    {
      "epoch": 0.02731114504803691,
      "grad_norm": 1247.5994873046875,
      "learning_rate": 1.3656565656565656e-05,
      "loss": 318.9619,
      "step": 3380
    },
    {
      "epoch": 0.027391947252321042,
      "grad_norm": 1123.76220703125,
      "learning_rate": 1.3696969696969697e-05,
      "loss": 235.3794,
      "step": 3390
    },
    {
      "epoch": 0.027472749456605175,
      "grad_norm": 2435.860107421875,
      "learning_rate": 1.3737373737373737e-05,
      "loss": 290.7648,
      "step": 3400
    },
    {
      "epoch": 0.027553551660889308,
      "grad_norm": 1895.08642578125,
      "learning_rate": 1.3777777777777778e-05,
      "loss": 228.9499,
      "step": 3410
    },
    {
      "epoch": 0.02763435386517344,
      "grad_norm": 1057.9578857421875,
      "learning_rate": 1.3818181818181818e-05,
      "loss": 199.0401,
      "step": 3420
    },
    {
      "epoch": 0.027715156069457576,
      "grad_norm": 1296.09130859375,
      "learning_rate": 1.385858585858586e-05,
      "loss": 238.6514,
      "step": 3430
    },
    {
      "epoch": 0.02779595827374171,
      "grad_norm": 3411.5341796875,
      "learning_rate": 1.3898989898989901e-05,
      "loss": 299.6798,
      "step": 3440
    },
    {
      "epoch": 0.02787676047802584,
      "grad_norm": 1198.8946533203125,
      "learning_rate": 1.3939393939393942e-05,
      "loss": 258.3361,
      "step": 3450
    },
    {
      "epoch": 0.027957562682309974,
      "grad_norm": 839.223388671875,
      "learning_rate": 1.3979797979797982e-05,
      "loss": 222.8733,
      "step": 3460
    },
    {
      "epoch": 0.028038364886594107,
      "grad_norm": 1462.1505126953125,
      "learning_rate": 1.402020202020202e-05,
      "loss": 206.9738,
      "step": 3470
    },
    {
      "epoch": 0.02811916709087824,
      "grad_norm": 1066.7890625,
      "learning_rate": 1.406060606060606e-05,
      "loss": 190.5506,
      "step": 3480
    },
    {
      "epoch": 0.028199969295162372,
      "grad_norm": 2753.951171875,
      "learning_rate": 1.41010101010101e-05,
      "loss": 210.7065,
      "step": 3490
    },
    {
      "epoch": 0.028280771499446505,
      "grad_norm": 1700.187744140625,
      "learning_rate": 1.4141414141414141e-05,
      "loss": 239.0851,
      "step": 3500
    },
    {
      "epoch": 0.028361573703730637,
      "grad_norm": 1913.965576171875,
      "learning_rate": 1.4181818181818181e-05,
      "loss": 244.5272,
      "step": 3510
    },
    {
      "epoch": 0.02844237590801477,
      "grad_norm": 1347.6934814453125,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 219.5849,
      "step": 3520
    },
    {
      "epoch": 0.028523178112298903,
      "grad_norm": 894.3698120117188,
      "learning_rate": 1.4262626262626264e-05,
      "loss": 261.3107,
      "step": 3530
    },
    {
      "epoch": 0.028603980316583035,
      "grad_norm": 649.42236328125,
      "learning_rate": 1.4303030303030305e-05,
      "loss": 202.5557,
      "step": 3540
    },
    {
      "epoch": 0.028684782520867168,
      "grad_norm": 824.8812255859375,
      "learning_rate": 1.4343434343434345e-05,
      "loss": 245.8003,
      "step": 3550
    },
    {
      "epoch": 0.0287655847251513,
      "grad_norm": 828.0931396484375,
      "learning_rate": 1.4383838383838386e-05,
      "loss": 260.2875,
      "step": 3560
    },
    {
      "epoch": 0.028846386929435437,
      "grad_norm": 1334.4947509765625,
      "learning_rate": 1.4424242424242426e-05,
      "loss": 232.7898,
      "step": 3570
    },
    {
      "epoch": 0.02892718913371957,
      "grad_norm": 1371.1171875,
      "learning_rate": 1.4464646464646464e-05,
      "loss": 418.4771,
      "step": 3580
    },
    {
      "epoch": 0.029007991338003702,
      "grad_norm": 18497.5234375,
      "learning_rate": 1.4505050505050504e-05,
      "loss": 303.3979,
      "step": 3590
    },
    {
      "epoch": 0.029088793542287834,
      "grad_norm": 1640.417724609375,
      "learning_rate": 1.4545454545454545e-05,
      "loss": 246.1203,
      "step": 3600
    },
    {
      "epoch": 0.029169595746571967,
      "grad_norm": 866.4635620117188,
      "learning_rate": 1.4585858585858587e-05,
      "loss": 227.0032,
      "step": 3610
    },
    {
      "epoch": 0.0292503979508561,
      "grad_norm": 1206.3389892578125,
      "learning_rate": 1.4626262626262627e-05,
      "loss": 240.7797,
      "step": 3620
    },
    {
      "epoch": 0.029331200155140232,
      "grad_norm": 1930.5679931640625,
      "learning_rate": 1.4666666666666668e-05,
      "loss": 244.9207,
      "step": 3630
    },
    {
      "epoch": 0.029412002359424365,
      "grad_norm": 1362.0755615234375,
      "learning_rate": 1.4707070707070709e-05,
      "loss": 223.5896,
      "step": 3640
    },
    {
      "epoch": 0.029492804563708497,
      "grad_norm": 1778.240478515625,
      "learning_rate": 1.4747474747474749e-05,
      "loss": 233.0804,
      "step": 3650
    },
    {
      "epoch": 0.02957360676799263,
      "grad_norm": 1185.7432861328125,
      "learning_rate": 1.478787878787879e-05,
      "loss": 269.5211,
      "step": 3660
    },
    {
      "epoch": 0.029654408972276763,
      "grad_norm": 1272.7274169921875,
      "learning_rate": 1.482828282828283e-05,
      "loss": 256.0854,
      "step": 3670
    },
    {
      "epoch": 0.029735211176560895,
      "grad_norm": 3724.482421875,
      "learning_rate": 1.486868686868687e-05,
      "loss": 220.0564,
      "step": 3680
    },
    {
      "epoch": 0.029816013380845028,
      "grad_norm": 1362.2408447265625,
      "learning_rate": 1.4909090909090908e-05,
      "loss": 196.9579,
      "step": 3690
    },
    {
      "epoch": 0.02989681558512916,
      "grad_norm": 1142.985107421875,
      "learning_rate": 1.494949494949495e-05,
      "loss": 298.3712,
      "step": 3700
    },
    {
      "epoch": 0.029977617789413297,
      "grad_norm": 1711.4461669921875,
      "learning_rate": 1.498989898989899e-05,
      "loss": 248.673,
      "step": 3710
    },
    {
      "epoch": 0.03005841999369743,
      "grad_norm": 1854.973876953125,
      "learning_rate": 1.5030303030303031e-05,
      "loss": 178.4528,
      "step": 3720
    },
    {
      "epoch": 0.030139222197981562,
      "grad_norm": 2415.3564453125,
      "learning_rate": 1.5070707070707072e-05,
      "loss": 279.0313,
      "step": 3730
    },
    {
      "epoch": 0.030220024402265695,
      "grad_norm": 1113.0447998046875,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 263.2642,
      "step": 3740
    },
    {
      "epoch": 0.030300826606549827,
      "grad_norm": 1523.1632080078125,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 292.6833,
      "step": 3750
    },
    {
      "epoch": 0.03038162881083396,
      "grad_norm": 1810.5382080078125,
      "learning_rate": 1.5191919191919193e-05,
      "loss": 260.2465,
      "step": 3760
    },
    {
      "epoch": 0.030462431015118092,
      "grad_norm": 2051.318115234375,
      "learning_rate": 1.5232323232323234e-05,
      "loss": 249.5686,
      "step": 3770
    },
    {
      "epoch": 0.030543233219402225,
      "grad_norm": 1145.482421875,
      "learning_rate": 1.5272727272727276e-05,
      "loss": 217.0,
      "step": 3780
    },
    {
      "epoch": 0.030624035423686358,
      "grad_norm": 1456.9969482421875,
      "learning_rate": 1.531313131313131e-05,
      "loss": 247.1355,
      "step": 3790
    },
    {
      "epoch": 0.03070483762797049,
      "grad_norm": 2063.9072265625,
      "learning_rate": 1.5353535353535354e-05,
      "loss": 317.8373,
      "step": 3800
    },
    {
      "epoch": 0.030785639832254623,
      "grad_norm": 1188.59130859375,
      "learning_rate": 1.5393939393939393e-05,
      "loss": 251.0659,
      "step": 3810
    },
    {
      "epoch": 0.030866442036538756,
      "grad_norm": 542.1653442382812,
      "learning_rate": 1.5434343434343435e-05,
      "loss": 205.6288,
      "step": 3820
    },
    {
      "epoch": 0.030947244240822888,
      "grad_norm": 858.66552734375,
      "learning_rate": 1.5474747474747474e-05,
      "loss": 261.1724,
      "step": 3830
    },
    {
      "epoch": 0.031028046445107024,
      "grad_norm": 1392.4208984375,
      "learning_rate": 1.5515151515151516e-05,
      "loss": 263.898,
      "step": 3840
    },
    {
      "epoch": 0.031108848649391157,
      "grad_norm": 1089.10888671875,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 263.4895,
      "step": 3850
    },
    {
      "epoch": 0.03118965085367529,
      "grad_norm": 1323.1083984375,
      "learning_rate": 1.5595959595959597e-05,
      "loss": 224.5914,
      "step": 3860
    },
    {
      "epoch": 0.03127045305795942,
      "grad_norm": 748.7206420898438,
      "learning_rate": 1.563636363636364e-05,
      "loss": 185.8181,
      "step": 3870
    },
    {
      "epoch": 0.031351255262243555,
      "grad_norm": 1530.072021484375,
      "learning_rate": 1.5676767676767678e-05,
      "loss": 295.1081,
      "step": 3880
    },
    {
      "epoch": 0.031432057466527684,
      "grad_norm": 1390.1978759765625,
      "learning_rate": 1.571717171717172e-05,
      "loss": 221.7333,
      "step": 3890
    },
    {
      "epoch": 0.03151285967081182,
      "grad_norm": 1188.6934814453125,
      "learning_rate": 1.5757575757575756e-05,
      "loss": 231.4922,
      "step": 3900
    },
    {
      "epoch": 0.031593661875095956,
      "grad_norm": 1810.8616943359375,
      "learning_rate": 1.5797979797979798e-05,
      "loss": 226.2008,
      "step": 3910
    },
    {
      "epoch": 0.031674464079380085,
      "grad_norm": 1351.2021484375,
      "learning_rate": 1.5838383838383837e-05,
      "loss": 213.8082,
      "step": 3920
    },
    {
      "epoch": 0.03175526628366422,
      "grad_norm": 1504.8511962890625,
      "learning_rate": 1.587878787878788e-05,
      "loss": 237.731,
      "step": 3930
    },
    {
      "epoch": 0.03183606848794835,
      "grad_norm": 3990.205810546875,
      "learning_rate": 1.5919191919191918e-05,
      "loss": 308.5875,
      "step": 3940
    },
    {
      "epoch": 0.03191687069223249,
      "grad_norm": 1052.4140625,
      "learning_rate": 1.595959595959596e-05,
      "loss": 173.6135,
      "step": 3950
    },
    {
      "epoch": 0.031997672896516616,
      "grad_norm": 818.5986328125,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 255.928,
      "step": 3960
    },
    {
      "epoch": 0.03207847510080075,
      "grad_norm": 3015.482666015625,
      "learning_rate": 1.604040404040404e-05,
      "loss": 246.6157,
      "step": 3970
    },
    {
      "epoch": 0.03215927730508488,
      "grad_norm": 1520.350341796875,
      "learning_rate": 1.6080808080808084e-05,
      "loss": 294.5478,
      "step": 3980
    },
    {
      "epoch": 0.03224007950936902,
      "grad_norm": 1362.8385009765625,
      "learning_rate": 1.6121212121212123e-05,
      "loss": 263.2382,
      "step": 3990
    },
    {
      "epoch": 0.032320881713653146,
      "grad_norm": 1330.2135009765625,
      "learning_rate": 1.6161616161616165e-05,
      "loss": 227.9952,
      "step": 4000
    },
    {
      "epoch": 0.03240168391793728,
      "grad_norm": 2001.479248046875,
      "learning_rate": 1.62020202020202e-05,
      "loss": 373.7298,
      "step": 4010
    },
    {
      "epoch": 0.03248248612222141,
      "grad_norm": 670.8789672851562,
      "learning_rate": 1.6242424242424243e-05,
      "loss": 252.2481,
      "step": 4020
    },
    {
      "epoch": 0.03256328832650555,
      "grad_norm": 1504.35205078125,
      "learning_rate": 1.628282828282828e-05,
      "loss": 259.7328,
      "step": 4030
    },
    {
      "epoch": 0.03264409053078968,
      "grad_norm": 1177.47509765625,
      "learning_rate": 1.6323232323232324e-05,
      "loss": 220.6592,
      "step": 4040
    },
    {
      "epoch": 0.03272489273507381,
      "grad_norm": 889.9537353515625,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 210.0868,
      "step": 4050
    },
    {
      "epoch": 0.03280569493935795,
      "grad_norm": 1655.767333984375,
      "learning_rate": 1.6404040404040405e-05,
      "loss": 247.7082,
      "step": 4060
    },
    {
      "epoch": 0.03288649714364208,
      "grad_norm": 1741.26416015625,
      "learning_rate": 1.6444444444444447e-05,
      "loss": 213.1305,
      "step": 4070
    },
    {
      "epoch": 0.032967299347926214,
      "grad_norm": 1701.3470458984375,
      "learning_rate": 1.6484848484848486e-05,
      "loss": 197.4172,
      "step": 4080
    },
    {
      "epoch": 0.03304810155221034,
      "grad_norm": 1241.48876953125,
      "learning_rate": 1.6525252525252528e-05,
      "loss": 200.895,
      "step": 4090
    },
    {
      "epoch": 0.03312890375649448,
      "grad_norm": 4305.5234375,
      "learning_rate": 1.6565656565656567e-05,
      "loss": 270.1561,
      "step": 4100
    },
    {
      "epoch": 0.03320970596077861,
      "grad_norm": 1233.9559326171875,
      "learning_rate": 1.6606060606060606e-05,
      "loss": 234.697,
      "step": 4110
    },
    {
      "epoch": 0.033290508165062745,
      "grad_norm": 1864.9722900390625,
      "learning_rate": 1.6646464646464645e-05,
      "loss": 207.7519,
      "step": 4120
    },
    {
      "epoch": 0.033371310369346874,
      "grad_norm": 696.45654296875,
      "learning_rate": 1.6686868686868687e-05,
      "loss": 260.9977,
      "step": 4130
    },
    {
      "epoch": 0.03345211257363101,
      "grad_norm": 1083.8914794921875,
      "learning_rate": 1.672727272727273e-05,
      "loss": 296.5648,
      "step": 4140
    },
    {
      "epoch": 0.03353291477791514,
      "grad_norm": 787.8980102539062,
      "learning_rate": 1.6767676767676768e-05,
      "loss": 252.0068,
      "step": 4150
    },
    {
      "epoch": 0.033613716982199275,
      "grad_norm": 3963.899658203125,
      "learning_rate": 1.680808080808081e-05,
      "loss": 239.0976,
      "step": 4160
    },
    {
      "epoch": 0.033694519186483404,
      "grad_norm": 1345.8841552734375,
      "learning_rate": 1.684848484848485e-05,
      "loss": 200.5568,
      "step": 4170
    },
    {
      "epoch": 0.03377532139076754,
      "grad_norm": 1667.1441650390625,
      "learning_rate": 1.688888888888889e-05,
      "loss": 246.528,
      "step": 4180
    },
    {
      "epoch": 0.033856123595051676,
      "grad_norm": 941.3829956054688,
      "learning_rate": 1.692929292929293e-05,
      "loss": 188.0032,
      "step": 4190
    },
    {
      "epoch": 0.033936925799335806,
      "grad_norm": 2224.00048828125,
      "learning_rate": 1.6969696969696972e-05,
      "loss": 233.2688,
      "step": 4200
    },
    {
      "epoch": 0.03401772800361994,
      "grad_norm": 990.577880859375,
      "learning_rate": 1.701010101010101e-05,
      "loss": 229.0408,
      "step": 4210
    },
    {
      "epoch": 0.03409853020790407,
      "grad_norm": 1741.591064453125,
      "learning_rate": 1.705050505050505e-05,
      "loss": 210.8973,
      "step": 4220
    },
    {
      "epoch": 0.03417933241218821,
      "grad_norm": 1565.2149658203125,
      "learning_rate": 1.7090909090909092e-05,
      "loss": 172.9691,
      "step": 4230
    },
    {
      "epoch": 0.034260134616472336,
      "grad_norm": 1411.6668701171875,
      "learning_rate": 1.713131313131313e-05,
      "loss": 223.8018,
      "step": 4240
    },
    {
      "epoch": 0.03434093682075647,
      "grad_norm": 849.447998046875,
      "learning_rate": 1.7171717171717173e-05,
      "loss": 277.72,
      "step": 4250
    },
    {
      "epoch": 0.0344217390250406,
      "grad_norm": 1456.3353271484375,
      "learning_rate": 1.7212121212121212e-05,
      "loss": 269.1795,
      "step": 4260
    },
    {
      "epoch": 0.03450254122932474,
      "grad_norm": 2039.048583984375,
      "learning_rate": 1.7252525252525255e-05,
      "loss": 203.6644,
      "step": 4270
    },
    {
      "epoch": 0.03458334343360887,
      "grad_norm": 1037.1063232421875,
      "learning_rate": 1.7292929292929293e-05,
      "loss": 268.1442,
      "step": 4280
    },
    {
      "epoch": 0.034664145637893,
      "grad_norm": 1481.98095703125,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 246.1609,
      "step": 4290
    },
    {
      "epoch": 0.03474494784217713,
      "grad_norm": 1042.147216796875,
      "learning_rate": 1.7373737373737375e-05,
      "loss": 360.0711,
      "step": 4300
    },
    {
      "epoch": 0.03482575004646127,
      "grad_norm": 1008.8258666992188,
      "learning_rate": 1.7414141414141417e-05,
      "loss": 254.1684,
      "step": 4310
    },
    {
      "epoch": 0.0349065522507454,
      "grad_norm": 1818.73681640625,
      "learning_rate": 1.7454545454545456e-05,
      "loss": 248.8469,
      "step": 4320
    },
    {
      "epoch": 0.03498735445502953,
      "grad_norm": 2598.832763671875,
      "learning_rate": 1.7494949494949494e-05,
      "loss": 215.4962,
      "step": 4330
    },
    {
      "epoch": 0.03506815665931367,
      "grad_norm": 5505.1572265625,
      "learning_rate": 1.7535353535353537e-05,
      "loss": 206.1825,
      "step": 4340
    },
    {
      "epoch": 0.0351489588635978,
      "grad_norm": 872.9111328125,
      "learning_rate": 1.7575757575757576e-05,
      "loss": 226.9096,
      "step": 4350
    },
    {
      "epoch": 0.035229761067881935,
      "grad_norm": 1309.483154296875,
      "learning_rate": 1.7616161616161618e-05,
      "loss": 347.4825,
      "step": 4360
    },
    {
      "epoch": 0.035310563272166064,
      "grad_norm": 1847.357666015625,
      "learning_rate": 1.7656565656565657e-05,
      "loss": 283.7126,
      "step": 4370
    },
    {
      "epoch": 0.0353913654764502,
      "grad_norm": 1132.7510986328125,
      "learning_rate": 1.76969696969697e-05,
      "loss": 238.7522,
      "step": 4380
    },
    {
      "epoch": 0.03547216768073433,
      "grad_norm": 1338.4906005859375,
      "learning_rate": 1.7737373737373738e-05,
      "loss": 206.4677,
      "step": 4390
    },
    {
      "epoch": 0.035552969885018465,
      "grad_norm": 889.9144897460938,
      "learning_rate": 1.777777777777778e-05,
      "loss": 218.258,
      "step": 4400
    },
    {
      "epoch": 0.035633772089302594,
      "grad_norm": 1081.747314453125,
      "learning_rate": 1.781818181818182e-05,
      "loss": 227.4267,
      "step": 4410
    },
    {
      "epoch": 0.03571457429358673,
      "grad_norm": 1337.2747802734375,
      "learning_rate": 1.785858585858586e-05,
      "loss": 216.5905,
      "step": 4420
    },
    {
      "epoch": 0.03579537649787086,
      "grad_norm": 1070.0733642578125,
      "learning_rate": 1.78989898989899e-05,
      "loss": 244.2413,
      "step": 4430
    },
    {
      "epoch": 0.035876178702154995,
      "grad_norm": 2713.52392578125,
      "learning_rate": 1.793939393939394e-05,
      "loss": 214.338,
      "step": 4440
    },
    {
      "epoch": 0.035956980906439125,
      "grad_norm": 1579.244873046875,
      "learning_rate": 1.797979797979798e-05,
      "loss": 236.9733,
      "step": 4450
    },
    {
      "epoch": 0.03603778311072326,
      "grad_norm": 1429.0421142578125,
      "learning_rate": 1.802020202020202e-05,
      "loss": 210.1912,
      "step": 4460
    },
    {
      "epoch": 0.0361185853150074,
      "grad_norm": 1236.484375,
      "learning_rate": 1.8060606060606062e-05,
      "loss": 209.4003,
      "step": 4470
    },
    {
      "epoch": 0.036199387519291526,
      "grad_norm": 2510.634521484375,
      "learning_rate": 1.81010101010101e-05,
      "loss": 261.3127,
      "step": 4480
    },
    {
      "epoch": 0.03628018972357566,
      "grad_norm": 697.04345703125,
      "learning_rate": 1.8141414141414143e-05,
      "loss": 202.3308,
      "step": 4490
    },
    {
      "epoch": 0.03636099192785979,
      "grad_norm": 1664.605712890625,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 230.7549,
      "step": 4500
    },
    {
      "epoch": 0.03644179413214393,
      "grad_norm": 1968.6279296875,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 222.0125,
      "step": 4510
    },
    {
      "epoch": 0.036522596336428056,
      "grad_norm": 1813.247314453125,
      "learning_rate": 1.8262626262626263e-05,
      "loss": 206.1146,
      "step": 4520
    },
    {
      "epoch": 0.03660339854071219,
      "grad_norm": 1681.3162841796875,
      "learning_rate": 1.8303030303030305e-05,
      "loss": 281.5203,
      "step": 4530
    },
    {
      "epoch": 0.03668420074499632,
      "grad_norm": 813.0327758789062,
      "learning_rate": 1.8343434343434344e-05,
      "loss": 241.7396,
      "step": 4540
    },
    {
      "epoch": 0.03676500294928046,
      "grad_norm": 1714.4927978515625,
      "learning_rate": 1.8383838383838383e-05,
      "loss": 229.0337,
      "step": 4550
    },
    {
      "epoch": 0.03684580515356459,
      "grad_norm": 1173.26318359375,
      "learning_rate": 1.8424242424242425e-05,
      "loss": 167.8814,
      "step": 4560
    },
    {
      "epoch": 0.03692660735784872,
      "grad_norm": 1044.22509765625,
      "learning_rate": 1.8464646464646464e-05,
      "loss": 181.4134,
      "step": 4570
    },
    {
      "epoch": 0.03700740956213285,
      "grad_norm": 1544.4964599609375,
      "learning_rate": 1.8505050505050506e-05,
      "loss": 264.1711,
      "step": 4580
    },
    {
      "epoch": 0.03708821176641699,
      "grad_norm": 3204.8271484375,
      "learning_rate": 1.8545454545454545e-05,
      "loss": 209.0515,
      "step": 4590
    },
    {
      "epoch": 0.03716901397070112,
      "grad_norm": 1948.9998779296875,
      "learning_rate": 1.8585858585858588e-05,
      "loss": 204.1481,
      "step": 4600
    },
    {
      "epoch": 0.037249816174985254,
      "grad_norm": 985.3388671875,
      "learning_rate": 1.8626262626262626e-05,
      "loss": 251.0652,
      "step": 4610
    },
    {
      "epoch": 0.03733061837926939,
      "grad_norm": 4716.29833984375,
      "learning_rate": 1.866666666666667e-05,
      "loss": 234.005,
      "step": 4620
    },
    {
      "epoch": 0.03741142058355352,
      "grad_norm": 2745.129150390625,
      "learning_rate": 1.8707070707070707e-05,
      "loss": 222.8053,
      "step": 4630
    },
    {
      "epoch": 0.037492222787837655,
      "grad_norm": 852.2494506835938,
      "learning_rate": 1.874747474747475e-05,
      "loss": 244.6,
      "step": 4640
    },
    {
      "epoch": 0.037573024992121784,
      "grad_norm": 1276.906494140625,
      "learning_rate": 1.878787878787879e-05,
      "loss": 243.4739,
      "step": 4650
    },
    {
      "epoch": 0.03765382719640592,
      "grad_norm": 2488.490478515625,
      "learning_rate": 1.8828282828282827e-05,
      "loss": 241.5105,
      "step": 4660
    },
    {
      "epoch": 0.03773462940069005,
      "grad_norm": 1208.5731201171875,
      "learning_rate": 1.886868686868687e-05,
      "loss": 266.4298,
      "step": 4670
    },
    {
      "epoch": 0.037815431604974185,
      "grad_norm": 1110.9935302734375,
      "learning_rate": 1.890909090909091e-05,
      "loss": 220.2013,
      "step": 4680
    },
    {
      "epoch": 0.037896233809258315,
      "grad_norm": 966.4763793945312,
      "learning_rate": 1.894949494949495e-05,
      "loss": 213.4089,
      "step": 4690
    },
    {
      "epoch": 0.03797703601354245,
      "grad_norm": 888.4136352539062,
      "learning_rate": 1.898989898989899e-05,
      "loss": 192.6133,
      "step": 4700
    },
    {
      "epoch": 0.03805783821782658,
      "grad_norm": 1441.930419921875,
      "learning_rate": 1.9030303030303032e-05,
      "loss": 210.6855,
      "step": 4710
    },
    {
      "epoch": 0.038138640422110716,
      "grad_norm": 1268.2919921875,
      "learning_rate": 1.907070707070707e-05,
      "loss": 196.9399,
      "step": 4720
    },
    {
      "epoch": 0.038219442626394845,
      "grad_norm": 714.101318359375,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 236.1493,
      "step": 4730
    },
    {
      "epoch": 0.03830024483067898,
      "grad_norm": 1360.3662109375,
      "learning_rate": 1.9151515151515155e-05,
      "loss": 277.1614,
      "step": 4740
    },
    {
      "epoch": 0.03838104703496312,
      "grad_norm": 857.1802368164062,
      "learning_rate": 1.919191919191919e-05,
      "loss": 233.6975,
      "step": 4750
    },
    {
      "epoch": 0.038461849239247246,
      "grad_norm": 1430.3370361328125,
      "learning_rate": 1.9232323232323233e-05,
      "loss": 206.9375,
      "step": 4760
    },
    {
      "epoch": 0.03854265144353138,
      "grad_norm": 999.745849609375,
      "learning_rate": 1.9272727272727272e-05,
      "loss": 177.6682,
      "step": 4770
    },
    {
      "epoch": 0.03862345364781551,
      "grad_norm": 1979.0234375,
      "learning_rate": 1.9313131313131314e-05,
      "loss": 237.5471,
      "step": 4780
    },
    {
      "epoch": 0.03870425585209965,
      "grad_norm": 1399.9544677734375,
      "learning_rate": 1.9353535353535353e-05,
      "loss": 209.8267,
      "step": 4790
    },
    {
      "epoch": 0.03878505805638378,
      "grad_norm": 1058.5128173828125,
      "learning_rate": 1.9393939393939395e-05,
      "loss": 206.1269,
      "step": 4800
    },
    {
      "epoch": 0.03886586026066791,
      "grad_norm": 1852.674072265625,
      "learning_rate": 1.9434343434343434e-05,
      "loss": 192.6013,
      "step": 4810
    },
    {
      "epoch": 0.03894666246495204,
      "grad_norm": 1104.2967529296875,
      "learning_rate": 1.9474747474747476e-05,
      "loss": 252.5522,
      "step": 4820
    },
    {
      "epoch": 0.03902746466923618,
      "grad_norm": 1426.0396728515625,
      "learning_rate": 1.951515151515152e-05,
      "loss": 250.6448,
      "step": 4830
    },
    {
      "epoch": 0.03910826687352031,
      "grad_norm": 1632.4510498046875,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 163.3638,
      "step": 4840
    },
    {
      "epoch": 0.03918906907780444,
      "grad_norm": 700.0907592773438,
      "learning_rate": 1.95959595959596e-05,
      "loss": 236.7388,
      "step": 4850
    },
    {
      "epoch": 0.03926987128208857,
      "grad_norm": 1205.572265625,
      "learning_rate": 1.9636363636363635e-05,
      "loss": 272.2705,
      "step": 4860
    },
    {
      "epoch": 0.03935067348637271,
      "grad_norm": 799.412353515625,
      "learning_rate": 1.9676767676767677e-05,
      "loss": 171.4291,
      "step": 4870
    },
    {
      "epoch": 0.03943147569065684,
      "grad_norm": 1350.2025146484375,
      "learning_rate": 1.9717171717171716e-05,
      "loss": 233.9921,
      "step": 4880
    },
    {
      "epoch": 0.039512277894940974,
      "grad_norm": 976.219970703125,
      "learning_rate": 1.975757575757576e-05,
      "loss": 189.0711,
      "step": 4890
    },
    {
      "epoch": 0.03959308009922511,
      "grad_norm": 947.8401489257812,
      "learning_rate": 1.9797979797979797e-05,
      "loss": 207.2786,
      "step": 4900
    },
    {
      "epoch": 0.03967388230350924,
      "grad_norm": 1402.2440185546875,
      "learning_rate": 1.983838383838384e-05,
      "loss": 233.6717,
      "step": 4910
    },
    {
      "epoch": 0.039754684507793375,
      "grad_norm": 2319.2314453125,
      "learning_rate": 1.987878787878788e-05,
      "loss": 268.4254,
      "step": 4920
    },
    {
      "epoch": 0.039835486712077504,
      "grad_norm": 1344.019775390625,
      "learning_rate": 1.991919191919192e-05,
      "loss": 215.5304,
      "step": 4930
    },
    {
      "epoch": 0.03991628891636164,
      "grad_norm": 1209.1622314453125,
      "learning_rate": 1.9959595959595963e-05,
      "loss": 202.8059,
      "step": 4940
    },
    {
      "epoch": 0.03999709112064577,
      "grad_norm": 1872.3892822265625,
      "learning_rate": 2e-05,
      "loss": 193.5764,
      "step": 4950
    },
    {
      "epoch": 0.040077893324929906,
      "grad_norm": 1944.2449951171875,
      "learning_rate": 2.0040404040404044e-05,
      "loss": 273.8487,
      "step": 4960
    },
    {
      "epoch": 0.040158695529214035,
      "grad_norm": 988.1495361328125,
      "learning_rate": 2.008080808080808e-05,
      "loss": 202.3245,
      "step": 4970
    },
    {
      "epoch": 0.04023949773349817,
      "grad_norm": 1082.6280517578125,
      "learning_rate": 2.012121212121212e-05,
      "loss": 190.6009,
      "step": 4980
    },
    {
      "epoch": 0.0403202999377823,
      "grad_norm": 1510.5738525390625,
      "learning_rate": 2.016161616161616e-05,
      "loss": 262.4282,
      "step": 4990
    },
    {
      "epoch": 0.040401102142066436,
      "grad_norm": 1080.0328369140625,
      "learning_rate": 2.0202020202020203e-05,
      "loss": 179.5178,
      "step": 5000
    },
    {
      "epoch": 0.040481904346350565,
      "grad_norm": 1204.5341796875,
      "learning_rate": 2.0242424242424245e-05,
      "loss": 208.4234,
      "step": 5010
    },
    {
      "epoch": 0.0405627065506347,
      "grad_norm": 788.6203002929688,
      "learning_rate": 2.0282828282828284e-05,
      "loss": 222.1854,
      "step": 5020
    },
    {
      "epoch": 0.04064350875491884,
      "grad_norm": 2447.934326171875,
      "learning_rate": 2.0323232323232326e-05,
      "loss": 183.1969,
      "step": 5030
    },
    {
      "epoch": 0.04072431095920297,
      "grad_norm": 1879.5914306640625,
      "learning_rate": 2.0363636363636365e-05,
      "loss": 235.8428,
      "step": 5040
    },
    {
      "epoch": 0.0408051131634871,
      "grad_norm": 859.5083618164062,
      "learning_rate": 2.0404040404040407e-05,
      "loss": 223.1974,
      "step": 5050
    },
    {
      "epoch": 0.04088591536777123,
      "grad_norm": 591.982421875,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 195.905,
      "step": 5060
    },
    {
      "epoch": 0.04096671757205537,
      "grad_norm": 2516.256103515625,
      "learning_rate": 2.0484848484848485e-05,
      "loss": 224.4586,
      "step": 5070
    },
    {
      "epoch": 0.0410475197763395,
      "grad_norm": 1155.78271484375,
      "learning_rate": 2.0525252525252524e-05,
      "loss": 237.8034,
      "step": 5080
    },
    {
      "epoch": 0.04112832198062363,
      "grad_norm": 760.8511962890625,
      "learning_rate": 2.0565656565656566e-05,
      "loss": 213.4372,
      "step": 5090
    },
    {
      "epoch": 0.04120912418490776,
      "grad_norm": 746.3182983398438,
      "learning_rate": 2.0606060606060608e-05,
      "loss": 246.9279,
      "step": 5100
    },
    {
      "epoch": 0.0412899263891919,
      "grad_norm": 1112.6119384765625,
      "learning_rate": 2.0646464646464647e-05,
      "loss": 215.4636,
      "step": 5110
    },
    {
      "epoch": 0.04137072859347603,
      "grad_norm": 1308.880126953125,
      "learning_rate": 2.068686868686869e-05,
      "loss": 184.3576,
      "step": 5120
    },
    {
      "epoch": 0.041451530797760164,
      "grad_norm": 1182.3695068359375,
      "learning_rate": 2.0727272727272728e-05,
      "loss": 251.3663,
      "step": 5130
    },
    {
      "epoch": 0.04153233300204429,
      "grad_norm": 3545.449951171875,
      "learning_rate": 2.076767676767677e-05,
      "loss": 221.7183,
      "step": 5140
    },
    {
      "epoch": 0.04161313520632843,
      "grad_norm": 1155.616455078125,
      "learning_rate": 2.080808080808081e-05,
      "loss": 181.7703,
      "step": 5150
    },
    {
      "epoch": 0.04169393741061256,
      "grad_norm": 927.0892333984375,
      "learning_rate": 2.084848484848485e-05,
      "loss": 242.7771,
      "step": 5160
    },
    {
      "epoch": 0.041774739614896694,
      "grad_norm": 1621.09326171875,
      "learning_rate": 2.088888888888889e-05,
      "loss": 168.8398,
      "step": 5170
    },
    {
      "epoch": 0.04185554181918083,
      "grad_norm": 1823.0281982421875,
      "learning_rate": 2.092929292929293e-05,
      "loss": 226.3993,
      "step": 5180
    },
    {
      "epoch": 0.04193634402346496,
      "grad_norm": 1904.581298828125,
      "learning_rate": 2.096969696969697e-05,
      "loss": 274.6904,
      "step": 5190
    },
    {
      "epoch": 0.042017146227749096,
      "grad_norm": 1195.8973388671875,
      "learning_rate": 2.101010101010101e-05,
      "loss": 193.929,
      "step": 5200
    },
    {
      "epoch": 0.042097948432033225,
      "grad_norm": 809.5712890625,
      "learning_rate": 2.1050505050505052e-05,
      "loss": 183.259,
      "step": 5210
    },
    {
      "epoch": 0.04217875063631736,
      "grad_norm": 1392.5491943359375,
      "learning_rate": 2.109090909090909e-05,
      "loss": 220.0326,
      "step": 5220
    },
    {
      "epoch": 0.04225955284060149,
      "grad_norm": 1818.6051025390625,
      "learning_rate": 2.1131313131313134e-05,
      "loss": 209.3423,
      "step": 5230
    },
    {
      "epoch": 0.042340355044885626,
      "grad_norm": 756.583740234375,
      "learning_rate": 2.1171717171717172e-05,
      "loss": 152.79,
      "step": 5240
    },
    {
      "epoch": 0.042421157249169755,
      "grad_norm": 1358.5194091796875,
      "learning_rate": 2.1212121212121215e-05,
      "loss": 223.6846,
      "step": 5250
    },
    {
      "epoch": 0.04250195945345389,
      "grad_norm": 2302.727783203125,
      "learning_rate": 2.1252525252525254e-05,
      "loss": 206.5412,
      "step": 5260
    },
    {
      "epoch": 0.04258276165773802,
      "grad_norm": 1090.666259765625,
      "learning_rate": 2.1292929292929296e-05,
      "loss": 197.9379,
      "step": 5270
    },
    {
      "epoch": 0.04266356386202216,
      "grad_norm": 1535.5264892578125,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 172.5529,
      "step": 5280
    },
    {
      "epoch": 0.042744366066306286,
      "grad_norm": 1242.1055908203125,
      "learning_rate": 2.1373737373737373e-05,
      "loss": 182.2667,
      "step": 5290
    },
    {
      "epoch": 0.04282516827059042,
      "grad_norm": 1571.221923828125,
      "learning_rate": 2.1414141414141416e-05,
      "loss": 206.9152,
      "step": 5300
    },
    {
      "epoch": 0.04290597047487456,
      "grad_norm": 1733.92578125,
      "learning_rate": 2.1454545454545455e-05,
      "loss": 253.6228,
      "step": 5310
    },
    {
      "epoch": 0.04298677267915869,
      "grad_norm": 1736.4722900390625,
      "learning_rate": 2.1494949494949497e-05,
      "loss": 209.5105,
      "step": 5320
    },
    {
      "epoch": 0.04306757488344282,
      "grad_norm": 846.6854248046875,
      "learning_rate": 2.1535353535353536e-05,
      "loss": 227.6331,
      "step": 5330
    },
    {
      "epoch": 0.04314837708772695,
      "grad_norm": 793.491943359375,
      "learning_rate": 2.1575757575757578e-05,
      "loss": 190.9206,
      "step": 5340
    },
    {
      "epoch": 0.04322917929201109,
      "grad_norm": 1314.4940185546875,
      "learning_rate": 2.1616161616161617e-05,
      "loss": 278.1586,
      "step": 5350
    },
    {
      "epoch": 0.04330998149629522,
      "grad_norm": 1807.1669921875,
      "learning_rate": 2.165656565656566e-05,
      "loss": 246.6954,
      "step": 5360
    },
    {
      "epoch": 0.043390783700579354,
      "grad_norm": 1456.6739501953125,
      "learning_rate": 2.1696969696969698e-05,
      "loss": 198.1051,
      "step": 5370
    },
    {
      "epoch": 0.04347158590486348,
      "grad_norm": 2645.863037109375,
      "learning_rate": 2.173737373737374e-05,
      "loss": 263.7012,
      "step": 5380
    },
    {
      "epoch": 0.04355238810914762,
      "grad_norm": 890.2818603515625,
      "learning_rate": 2.177777777777778e-05,
      "loss": 206.08,
      "step": 5390
    },
    {
      "epoch": 0.04363319031343175,
      "grad_norm": 1066.948974609375,
      "learning_rate": 2.1818181818181818e-05,
      "loss": 203.3024,
      "step": 5400
    },
    {
      "epoch": 0.043713992517715884,
      "grad_norm": 1678.3651123046875,
      "learning_rate": 2.185858585858586e-05,
      "loss": 287.4994,
      "step": 5410
    },
    {
      "epoch": 0.04379479472200001,
      "grad_norm": 1427.133544921875,
      "learning_rate": 2.18989898989899e-05,
      "loss": 236.3808,
      "step": 5420
    },
    {
      "epoch": 0.04387559692628415,
      "grad_norm": 993.3723754882812,
      "learning_rate": 2.193939393939394e-05,
      "loss": 221.2247,
      "step": 5430
    },
    {
      "epoch": 0.04395639913056828,
      "grad_norm": 919.2279663085938,
      "learning_rate": 2.197979797979798e-05,
      "loss": 232.8961,
      "step": 5440
    },
    {
      "epoch": 0.044037201334852415,
      "grad_norm": 1196.51904296875,
      "learning_rate": 2.2020202020202022e-05,
      "loss": 208.7773,
      "step": 5450
    },
    {
      "epoch": 0.04411800353913655,
      "grad_norm": 937.6903076171875,
      "learning_rate": 2.206060606060606e-05,
      "loss": 159.7425,
      "step": 5460
    },
    {
      "epoch": 0.04419880574342068,
      "grad_norm": 2946.419921875,
      "learning_rate": 2.2101010101010103e-05,
      "loss": 201.0844,
      "step": 5470
    },
    {
      "epoch": 0.044279607947704816,
      "grad_norm": 1663.4422607421875,
      "learning_rate": 2.2141414141414142e-05,
      "loss": 140.8333,
      "step": 5480
    },
    {
      "epoch": 0.044360410151988945,
      "grad_norm": 1202.589599609375,
      "learning_rate": 2.2181818181818184e-05,
      "loss": 210.6169,
      "step": 5490
    },
    {
      "epoch": 0.04444121235627308,
      "grad_norm": 1676.0555419921875,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 300.5972,
      "step": 5500
    },
    {
      "epoch": 0.04452201456055721,
      "grad_norm": 1122.7333984375,
      "learning_rate": 2.2262626262626262e-05,
      "loss": 223.7688,
      "step": 5510
    },
    {
      "epoch": 0.044602816764841346,
      "grad_norm": 842.3754272460938,
      "learning_rate": 2.2303030303030304e-05,
      "loss": 231.1573,
      "step": 5520
    },
    {
      "epoch": 0.044683618969125476,
      "grad_norm": 912.3519897460938,
      "learning_rate": 2.2343434343434343e-05,
      "loss": 161.5479,
      "step": 5530
    },
    {
      "epoch": 0.04476442117340961,
      "grad_norm": 2117.377197265625,
      "learning_rate": 2.2383838383838385e-05,
      "loss": 178.251,
      "step": 5540
    },
    {
      "epoch": 0.04484522337769374,
      "grad_norm": 1402.2164306640625,
      "learning_rate": 2.2424242424242424e-05,
      "loss": 209.2086,
      "step": 5550
    },
    {
      "epoch": 0.04492602558197788,
      "grad_norm": 1458.323974609375,
      "learning_rate": 2.2464646464646467e-05,
      "loss": 243.9479,
      "step": 5560
    },
    {
      "epoch": 0.045006827786262006,
      "grad_norm": 2175.216796875,
      "learning_rate": 2.2505050505050505e-05,
      "loss": 189.8892,
      "step": 5570
    },
    {
      "epoch": 0.04508762999054614,
      "grad_norm": 1899.4354248046875,
      "learning_rate": 2.2545454545454548e-05,
      "loss": 335.0552,
      "step": 5580
    },
    {
      "epoch": 0.04516843219483028,
      "grad_norm": 1230.814697265625,
      "learning_rate": 2.2585858585858587e-05,
      "loss": 194.9335,
      "step": 5590
    },
    {
      "epoch": 0.04524923439911441,
      "grad_norm": 2101.527587890625,
      "learning_rate": 2.262626262626263e-05,
      "loss": 257.3806,
      "step": 5600
    },
    {
      "epoch": 0.045330036603398544,
      "grad_norm": 1695.30810546875,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 219.7137,
      "step": 5610
    },
    {
      "epoch": 0.04541083880768267,
      "grad_norm": 1386.2855224609375,
      "learning_rate": 2.2707070707070706e-05,
      "loss": 236.2214,
      "step": 5620
    },
    {
      "epoch": 0.04549164101196681,
      "grad_norm": 1138.779052734375,
      "learning_rate": 2.274747474747475e-05,
      "loss": 192.9845,
      "step": 5630
    },
    {
      "epoch": 0.04557244321625094,
      "grad_norm": 2650.991943359375,
      "learning_rate": 2.2787878787878788e-05,
      "loss": 233.3904,
      "step": 5640
    },
    {
      "epoch": 0.045653245420535074,
      "grad_norm": 1309.0333251953125,
      "learning_rate": 2.282828282828283e-05,
      "loss": 225.3846,
      "step": 5650
    },
    {
      "epoch": 0.0457340476248192,
      "grad_norm": 930.385009765625,
      "learning_rate": 2.286868686868687e-05,
      "loss": 236.1336,
      "step": 5660
    },
    {
      "epoch": 0.04581484982910334,
      "grad_norm": 1646.2891845703125,
      "learning_rate": 2.290909090909091e-05,
      "loss": 227.3526,
      "step": 5670
    },
    {
      "epoch": 0.04589565203338747,
      "grad_norm": 2285.751708984375,
      "learning_rate": 2.294949494949495e-05,
      "loss": 236.6346,
      "step": 5680
    },
    {
      "epoch": 0.045976454237671605,
      "grad_norm": 3180.75537109375,
      "learning_rate": 2.2989898989898992e-05,
      "loss": 177.5457,
      "step": 5690
    },
    {
      "epoch": 0.046057256441955734,
      "grad_norm": 1423.35009765625,
      "learning_rate": 2.3030303030303034e-05,
      "loss": 194.2139,
      "step": 5700
    },
    {
      "epoch": 0.04613805864623987,
      "grad_norm": 1577.701171875,
      "learning_rate": 2.307070707070707e-05,
      "loss": 183.8717,
      "step": 5710
    },
    {
      "epoch": 0.046218860850524,
      "grad_norm": 1255.1485595703125,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 213.8492,
      "step": 5720
    },
    {
      "epoch": 0.046299663054808135,
      "grad_norm": 1154.9453125,
      "learning_rate": 2.315151515151515e-05,
      "loss": 219.6154,
      "step": 5730
    },
    {
      "epoch": 0.04638046525909227,
      "grad_norm": 3208.9140625,
      "learning_rate": 2.3191919191919193e-05,
      "loss": 212.2527,
      "step": 5740
    },
    {
      "epoch": 0.0464612674633764,
      "grad_norm": 826.8831787109375,
      "learning_rate": 2.3232323232323232e-05,
      "loss": 193.0573,
      "step": 5750
    },
    {
      "epoch": 0.046542069667660536,
      "grad_norm": 953.578369140625,
      "learning_rate": 2.3272727272727274e-05,
      "loss": 200.1285,
      "step": 5760
    },
    {
      "epoch": 0.046622871871944666,
      "grad_norm": 948.6517944335938,
      "learning_rate": 2.3313131313131313e-05,
      "loss": 226.3946,
      "step": 5770
    },
    {
      "epoch": 0.0467036740762288,
      "grad_norm": 1502.9415283203125,
      "learning_rate": 2.3353535353535355e-05,
      "loss": 301.4247,
      "step": 5780
    },
    {
      "epoch": 0.04678447628051293,
      "grad_norm": 592.7190551757812,
      "learning_rate": 2.3393939393939397e-05,
      "loss": 171.6613,
      "step": 5790
    },
    {
      "epoch": 0.04686527848479707,
      "grad_norm": 774.3163452148438,
      "learning_rate": 2.3434343434343436e-05,
      "loss": 174.7567,
      "step": 5800
    },
    {
      "epoch": 0.046946080689081196,
      "grad_norm": 1000.3840942382812,
      "learning_rate": 2.347474747474748e-05,
      "loss": 140.1143,
      "step": 5810
    },
    {
      "epoch": 0.04702688289336533,
      "grad_norm": 1050.761474609375,
      "learning_rate": 2.3515151515151514e-05,
      "loss": 234.2542,
      "step": 5820
    },
    {
      "epoch": 0.04710768509764946,
      "grad_norm": 1076.979248046875,
      "learning_rate": 2.3555555555555556e-05,
      "loss": 170.2877,
      "step": 5830
    },
    {
      "epoch": 0.0471884873019336,
      "grad_norm": 1528.865478515625,
      "learning_rate": 2.3595959595959595e-05,
      "loss": 280.3715,
      "step": 5840
    },
    {
      "epoch": 0.047269289506217727,
      "grad_norm": 1554.0205078125,
      "learning_rate": 2.3636363636363637e-05,
      "loss": 258.9206,
      "step": 5850
    },
    {
      "epoch": 0.04735009171050186,
      "grad_norm": 969.7879028320312,
      "learning_rate": 2.3676767676767676e-05,
      "loss": 173.4592,
      "step": 5860
    },
    {
      "epoch": 0.047430893914786,
      "grad_norm": 1271.55322265625,
      "learning_rate": 2.371717171717172e-05,
      "loss": 187.7373,
      "step": 5870
    },
    {
      "epoch": 0.04751169611907013,
      "grad_norm": 757.3799438476562,
      "learning_rate": 2.375757575757576e-05,
      "loss": 206.0978,
      "step": 5880
    },
    {
      "epoch": 0.047592498323354264,
      "grad_norm": 1099.2119140625,
      "learning_rate": 2.37979797979798e-05,
      "loss": 191.4486,
      "step": 5890
    },
    {
      "epoch": 0.04767330052763839,
      "grad_norm": 895.0558471679688,
      "learning_rate": 2.3838383838383842e-05,
      "loss": 197.1677,
      "step": 5900
    },
    {
      "epoch": 0.04775410273192253,
      "grad_norm": 900.752685546875,
      "learning_rate": 2.387878787878788e-05,
      "loss": 209.3482,
      "step": 5910
    },
    {
      "epoch": 0.04783490493620666,
      "grad_norm": 865.3425903320312,
      "learning_rate": 2.3919191919191923e-05,
      "loss": 211.704,
      "step": 5920
    },
    {
      "epoch": 0.047915707140490794,
      "grad_norm": 1376.961181640625,
      "learning_rate": 2.395959595959596e-05,
      "loss": 197.2012,
      "step": 5930
    },
    {
      "epoch": 0.047996509344774924,
      "grad_norm": 2671.92236328125,
      "learning_rate": 2.4e-05,
      "loss": 262.8319,
      "step": 5940
    },
    {
      "epoch": 0.04807731154905906,
      "grad_norm": 4328.66552734375,
      "learning_rate": 2.404040404040404e-05,
      "loss": 263.4226,
      "step": 5950
    },
    {
      "epoch": 0.04815811375334319,
      "grad_norm": 1454.4398193359375,
      "learning_rate": 2.4080808080808082e-05,
      "loss": 173.7909,
      "step": 5960
    },
    {
      "epoch": 0.048238915957627325,
      "grad_norm": 1238.2913818359375,
      "learning_rate": 2.4121212121212124e-05,
      "loss": 190.8571,
      "step": 5970
    },
    {
      "epoch": 0.048319718161911454,
      "grad_norm": 1106.6146240234375,
      "learning_rate": 2.4161616161616163e-05,
      "loss": 252.6962,
      "step": 5980
    },
    {
      "epoch": 0.04840052036619559,
      "grad_norm": 1612.1171875,
      "learning_rate": 2.4202020202020205e-05,
      "loss": 176.1714,
      "step": 5990
    },
    {
      "epoch": 0.04848132257047972,
      "grad_norm": 684.4707641601562,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 236.6299,
      "step": 6000
    },
    {
      "epoch": 0.048562124774763855,
      "grad_norm": 5278.638671875,
      "learning_rate": 2.4282828282828286e-05,
      "loss": 200.9588,
      "step": 6010
    },
    {
      "epoch": 0.04864292697904799,
      "grad_norm": 2136.859375,
      "learning_rate": 2.4323232323232325e-05,
      "loss": 249.8048,
      "step": 6020
    },
    {
      "epoch": 0.04872372918333212,
      "grad_norm": 704.8456420898438,
      "learning_rate": 2.4363636363636364e-05,
      "loss": 210.816,
      "step": 6030
    },
    {
      "epoch": 0.04880453138761626,
      "grad_norm": 2405.291259765625,
      "learning_rate": 2.4404040404040403e-05,
      "loss": 180.7068,
      "step": 6040
    },
    {
      "epoch": 0.048885333591900386,
      "grad_norm": 1121.5928955078125,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 268.1764,
      "step": 6050
    },
    {
      "epoch": 0.04896613579618452,
      "grad_norm": 1185.4925537109375,
      "learning_rate": 2.4484848484848484e-05,
      "loss": 252.5901,
      "step": 6060
    },
    {
      "epoch": 0.04904693800046865,
      "grad_norm": 1037.7261962890625,
      "learning_rate": 2.4525252525252526e-05,
      "loss": 217.7089,
      "step": 6070
    },
    {
      "epoch": 0.04912774020475279,
      "grad_norm": 3574.91943359375,
      "learning_rate": 2.4565656565656568e-05,
      "loss": 248.0757,
      "step": 6080
    },
    {
      "epoch": 0.049208542409036916,
      "grad_norm": 1335.7510986328125,
      "learning_rate": 2.4606060606060607e-05,
      "loss": 243.5903,
      "step": 6090
    },
    {
      "epoch": 0.04928934461332105,
      "grad_norm": 1548.2281494140625,
      "learning_rate": 2.464646464646465e-05,
      "loss": 204.2808,
      "step": 6100
    },
    {
      "epoch": 0.04937014681760518,
      "grad_norm": 1327.641357421875,
      "learning_rate": 2.4686868686868688e-05,
      "loss": 175.3226,
      "step": 6110
    },
    {
      "epoch": 0.04945094902188932,
      "grad_norm": 1096.567626953125,
      "learning_rate": 2.472727272727273e-05,
      "loss": 251.3891,
      "step": 6120
    },
    {
      "epoch": 0.04953175122617345,
      "grad_norm": 916.0780639648438,
      "learning_rate": 2.476767676767677e-05,
      "loss": 265.5964,
      "step": 6130
    },
    {
      "epoch": 0.04961255343045758,
      "grad_norm": 3319.821533203125,
      "learning_rate": 2.4808080808080808e-05,
      "loss": 199.656,
      "step": 6140
    },
    {
      "epoch": 0.04969335563474172,
      "grad_norm": 804.5398559570312,
      "learning_rate": 2.4848484848484847e-05,
      "loss": 176.793,
      "step": 6150
    },
    {
      "epoch": 0.04977415783902585,
      "grad_norm": 1266.6590576171875,
      "learning_rate": 2.488888888888889e-05,
      "loss": 172.6065,
      "step": 6160
    },
    {
      "epoch": 0.049854960043309984,
      "grad_norm": 953.1856689453125,
      "learning_rate": 2.492929292929293e-05,
      "loss": 259.3056,
      "step": 6170
    },
    {
      "epoch": 0.049935762247594113,
      "grad_norm": 1643.679443359375,
      "learning_rate": 2.496969696969697e-05,
      "loss": 227.5671,
      "step": 6180
    },
    {
      "epoch": 0.05001656445187825,
      "grad_norm": 2092.3837890625,
      "learning_rate": 2.5010101010101013e-05,
      "loss": 231.7141,
      "step": 6190
    },
    {
      "epoch": 0.05009736665616238,
      "grad_norm": 5872.7822265625,
      "learning_rate": 2.505050505050505e-05,
      "loss": 307.3282,
      "step": 6200
    },
    {
      "epoch": 0.050178168860446515,
      "grad_norm": 1653.10888671875,
      "learning_rate": 2.5090909090909094e-05,
      "loss": 290.71,
      "step": 6210
    },
    {
      "epoch": 0.050258971064730644,
      "grad_norm": 5940.2861328125,
      "learning_rate": 2.5131313131313133e-05,
      "loss": 298.1718,
      "step": 6220
    },
    {
      "epoch": 0.05033977326901478,
      "grad_norm": 1056.6617431640625,
      "learning_rate": 2.5171717171717175e-05,
      "loss": 167.573,
      "step": 6230
    },
    {
      "epoch": 0.05042057547329891,
      "grad_norm": 1492.5479736328125,
      "learning_rate": 2.5212121212121214e-05,
      "loss": 209.9481,
      "step": 6240
    },
    {
      "epoch": 0.050501377677583045,
      "grad_norm": 764.5651245117188,
      "learning_rate": 2.5252525252525256e-05,
      "loss": 189.9945,
      "step": 6250
    },
    {
      "epoch": 0.050582179881867174,
      "grad_norm": 2933.18603515625,
      "learning_rate": 2.5292929292929295e-05,
      "loss": 228.2252,
      "step": 6260
    },
    {
      "epoch": 0.05066298208615131,
      "grad_norm": 2692.583740234375,
      "learning_rate": 2.5333333333333337e-05,
      "loss": 217.1123,
      "step": 6270
    },
    {
      "epoch": 0.05074378429043544,
      "grad_norm": 1611.5694580078125,
      "learning_rate": 2.5373737373737376e-05,
      "loss": 199.2745,
      "step": 6280
    },
    {
      "epoch": 0.050824586494719576,
      "grad_norm": 638.3251953125,
      "learning_rate": 2.5414141414141418e-05,
      "loss": 283.4336,
      "step": 6290
    },
    {
      "epoch": 0.05090538869900371,
      "grad_norm": 960.7551879882812,
      "learning_rate": 2.5454545454545454e-05,
      "loss": 199.2895,
      "step": 6300
    },
    {
      "epoch": 0.05098619090328784,
      "grad_norm": 1416.6865234375,
      "learning_rate": 2.5494949494949492e-05,
      "loss": 247.6437,
      "step": 6310
    },
    {
      "epoch": 0.05106699310757198,
      "grad_norm": 962.5587158203125,
      "learning_rate": 2.5535353535353535e-05,
      "loss": 222.514,
      "step": 6320
    },
    {
      "epoch": 0.051147795311856106,
      "grad_norm": 1019.0704956054688,
      "learning_rate": 2.5575757575757573e-05,
      "loss": 233.7968,
      "step": 6330
    },
    {
      "epoch": 0.05122859751614024,
      "grad_norm": 1380.1087646484375,
      "learning_rate": 2.5616161616161616e-05,
      "loss": 203.472,
      "step": 6340
    },
    {
      "epoch": 0.05130939972042437,
      "grad_norm": 765.1551513671875,
      "learning_rate": 2.5656565656565658e-05,
      "loss": 202.9591,
      "step": 6350
    },
    {
      "epoch": 0.05139020192470851,
      "grad_norm": 854.4512329101562,
      "learning_rate": 2.5696969696969697e-05,
      "loss": 152.8654,
      "step": 6360
    },
    {
      "epoch": 0.05147100412899264,
      "grad_norm": 1366.1529541015625,
      "learning_rate": 2.573737373737374e-05,
      "loss": 202.4912,
      "step": 6370
    },
    {
      "epoch": 0.05155180633327677,
      "grad_norm": 812.153564453125,
      "learning_rate": 2.5777777777777778e-05,
      "loss": 190.2283,
      "step": 6380
    },
    {
      "epoch": 0.0516326085375609,
      "grad_norm": 2072.30029296875,
      "learning_rate": 2.581818181818182e-05,
      "loss": 250.4601,
      "step": 6390
    },
    {
      "epoch": 0.05171341074184504,
      "grad_norm": 1064.25732421875,
      "learning_rate": 2.585858585858586e-05,
      "loss": 243.8253,
      "step": 6400
    },
    {
      "epoch": 0.05179421294612917,
      "grad_norm": 1004.585205078125,
      "learning_rate": 2.58989898989899e-05,
      "loss": 233.6981,
      "step": 6410
    },
    {
      "epoch": 0.0518750151504133,
      "grad_norm": 781.0443115234375,
      "learning_rate": 2.593939393939394e-05,
      "loss": 231.5708,
      "step": 6420
    },
    {
      "epoch": 0.05195581735469744,
      "grad_norm": 1038.6923828125,
      "learning_rate": 2.5979797979797982e-05,
      "loss": 166.9408,
      "step": 6430
    },
    {
      "epoch": 0.05203661955898157,
      "grad_norm": 1369.49560546875,
      "learning_rate": 2.602020202020202e-05,
      "loss": 212.7086,
      "step": 6440
    },
    {
      "epoch": 0.052117421763265705,
      "grad_norm": 1065.5115966796875,
      "learning_rate": 2.6060606060606063e-05,
      "loss": 197.232,
      "step": 6450
    },
    {
      "epoch": 0.052198223967549834,
      "grad_norm": 1192.5135498046875,
      "learning_rate": 2.6101010101010102e-05,
      "loss": 210.3559,
      "step": 6460
    },
    {
      "epoch": 0.05227902617183397,
      "grad_norm": 2817.4658203125,
      "learning_rate": 2.6141414141414145e-05,
      "loss": 218.5129,
      "step": 6470
    },
    {
      "epoch": 0.0523598283761181,
      "grad_norm": 1661.2547607421875,
      "learning_rate": 2.6181818181818187e-05,
      "loss": 251.3214,
      "step": 6480
    },
    {
      "epoch": 0.052440630580402235,
      "grad_norm": 1465.83251953125,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 167.4487,
      "step": 6490
    },
    {
      "epoch": 0.052521432784686364,
      "grad_norm": 1172.0814208984375,
      "learning_rate": 2.6262626262626268e-05,
      "loss": 202.9199,
      "step": 6500
    },
    {
      "epoch": 0.0526022349889705,
      "grad_norm": 845.3886108398438,
      "learning_rate": 2.63030303030303e-05,
      "loss": 245.5335,
      "step": 6510
    },
    {
      "epoch": 0.05268303719325463,
      "grad_norm": 1505.1903076171875,
      "learning_rate": 2.6343434343434342e-05,
      "loss": 189.6107,
      "step": 6520
    },
    {
      "epoch": 0.052763839397538766,
      "grad_norm": 855.8611450195312,
      "learning_rate": 2.6383838383838384e-05,
      "loss": 168.8279,
      "step": 6530
    },
    {
      "epoch": 0.052844641601822895,
      "grad_norm": 1719.1915283203125,
      "learning_rate": 2.6424242424242423e-05,
      "loss": 266.6213,
      "step": 6540
    },
    {
      "epoch": 0.05292544380610703,
      "grad_norm": 1334.455322265625,
      "learning_rate": 2.6464646464646466e-05,
      "loss": 154.8023,
      "step": 6550
    },
    {
      "epoch": 0.05300624601039116,
      "grad_norm": 1549.58154296875,
      "learning_rate": 2.6505050505050504e-05,
      "loss": 188.1264,
      "step": 6560
    },
    {
      "epoch": 0.053087048214675296,
      "grad_norm": 740.02587890625,
      "learning_rate": 2.6545454545454547e-05,
      "loss": 241.7192,
      "step": 6570
    },
    {
      "epoch": 0.05316785041895943,
      "grad_norm": 1759.15869140625,
      "learning_rate": 2.6585858585858585e-05,
      "loss": 249.789,
      "step": 6580
    },
    {
      "epoch": 0.05324865262324356,
      "grad_norm": 1615.3770751953125,
      "learning_rate": 2.6626262626262628e-05,
      "loss": 256.1343,
      "step": 6590
    },
    {
      "epoch": 0.0533294548275277,
      "grad_norm": 1187.103515625,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 153.0139,
      "step": 6600
    },
    {
      "epoch": 0.05341025703181183,
      "grad_norm": 835.982177734375,
      "learning_rate": 2.670707070707071e-05,
      "loss": 240.5858,
      "step": 6610
    },
    {
      "epoch": 0.05349105923609596,
      "grad_norm": 658.9365234375,
      "learning_rate": 2.6747474747474748e-05,
      "loss": 212.2474,
      "step": 6620
    },
    {
      "epoch": 0.05357186144038009,
      "grad_norm": 836.15185546875,
      "learning_rate": 2.678787878787879e-05,
      "loss": 203.4595,
      "step": 6630
    },
    {
      "epoch": 0.05365266364466423,
      "grad_norm": 1312.960205078125,
      "learning_rate": 2.682828282828283e-05,
      "loss": 178.5055,
      "step": 6640
    },
    {
      "epoch": 0.05373346584894836,
      "grad_norm": 2402.58642578125,
      "learning_rate": 2.686868686868687e-05,
      "loss": 204.5425,
      "step": 6650
    },
    {
      "epoch": 0.05381426805323249,
      "grad_norm": 743.2178344726562,
      "learning_rate": 2.6909090909090913e-05,
      "loss": 151.0765,
      "step": 6660
    },
    {
      "epoch": 0.05389507025751662,
      "grad_norm": 2009.14599609375,
      "learning_rate": 2.6949494949494952e-05,
      "loss": 255.5723,
      "step": 6670
    },
    {
      "epoch": 0.05397587246180076,
      "grad_norm": 1129.924560546875,
      "learning_rate": 2.6989898989898994e-05,
      "loss": 249.5275,
      "step": 6680
    },
    {
      "epoch": 0.05405667466608489,
      "grad_norm": 1877.5682373046875,
      "learning_rate": 2.7030303030303033e-05,
      "loss": 201.4787,
      "step": 6690
    },
    {
      "epoch": 0.054137476870369024,
      "grad_norm": 1205.5860595703125,
      "learning_rate": 2.7070707070707075e-05,
      "loss": 165.1917,
      "step": 6700
    },
    {
      "epoch": 0.05421827907465316,
      "grad_norm": 833.5079956054688,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 222.6354,
      "step": 6710
    },
    {
      "epoch": 0.05429908127893729,
      "grad_norm": 1644.57470703125,
      "learning_rate": 2.7151515151515157e-05,
      "loss": 183.0618,
      "step": 6720
    },
    {
      "epoch": 0.054379883483221425,
      "grad_norm": 1261.3482666015625,
      "learning_rate": 2.7191919191919192e-05,
      "loss": 204.8876,
      "step": 6730
    },
    {
      "epoch": 0.054460685687505554,
      "grad_norm": 1064.4910888671875,
      "learning_rate": 2.723232323232323e-05,
      "loss": 228.8735,
      "step": 6740
    },
    {
      "epoch": 0.05454148789178969,
      "grad_norm": 1227.28369140625,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 244.5206,
      "step": 6750
    },
    {
      "epoch": 0.05462229009607382,
      "grad_norm": 747.6671142578125,
      "learning_rate": 2.7313131313131312e-05,
      "loss": 171.5991,
      "step": 6760
    },
    {
      "epoch": 0.054703092300357956,
      "grad_norm": 1191.174560546875,
      "learning_rate": 2.7353535353535354e-05,
      "loss": 183.062,
      "step": 6770
    },
    {
      "epoch": 0.054783894504642085,
      "grad_norm": 1179.271484375,
      "learning_rate": 2.7393939393939393e-05,
      "loss": 203.4914,
      "step": 6780
    },
    {
      "epoch": 0.05486469670892622,
      "grad_norm": 1980.94287109375,
      "learning_rate": 2.7434343434343435e-05,
      "loss": 190.7682,
      "step": 6790
    },
    {
      "epoch": 0.05494549891321035,
      "grad_norm": 1313.760498046875,
      "learning_rate": 2.7474747474747474e-05,
      "loss": 179.4395,
      "step": 6800
    },
    {
      "epoch": 0.055026301117494486,
      "grad_norm": 818.7135620117188,
      "learning_rate": 2.7515151515151516e-05,
      "loss": 240.7207,
      "step": 6810
    },
    {
      "epoch": 0.055107103321778615,
      "grad_norm": 1303.9735107421875,
      "learning_rate": 2.7555555555555555e-05,
      "loss": 197.5866,
      "step": 6820
    },
    {
      "epoch": 0.05518790552606275,
      "grad_norm": 4817.638671875,
      "learning_rate": 2.7595959595959597e-05,
      "loss": 236.2139,
      "step": 6830
    },
    {
      "epoch": 0.05526870773034688,
      "grad_norm": 1369.7080078125,
      "learning_rate": 2.7636363636363636e-05,
      "loss": 155.043,
      "step": 6840
    },
    {
      "epoch": 0.05534950993463102,
      "grad_norm": 1351.29150390625,
      "learning_rate": 2.767676767676768e-05,
      "loss": 193.5722,
      "step": 6850
    },
    {
      "epoch": 0.05543031213891515,
      "grad_norm": 1340.113525390625,
      "learning_rate": 2.771717171717172e-05,
      "loss": 196.4928,
      "step": 6860
    },
    {
      "epoch": 0.05551111434319928,
      "grad_norm": 1829.1298828125,
      "learning_rate": 2.775757575757576e-05,
      "loss": 207.2559,
      "step": 6870
    },
    {
      "epoch": 0.05559191654748342,
      "grad_norm": 1614.317138671875,
      "learning_rate": 2.7797979797979802e-05,
      "loss": 191.8481,
      "step": 6880
    },
    {
      "epoch": 0.05567271875176755,
      "grad_norm": 1409.754150390625,
      "learning_rate": 2.783838383838384e-05,
      "loss": 172.1698,
      "step": 6890
    },
    {
      "epoch": 0.05575352095605168,
      "grad_norm": 1008.7220458984375,
      "learning_rate": 2.7878787878787883e-05,
      "loss": 204.197,
      "step": 6900
    },
    {
      "epoch": 0.05583432316033581,
      "grad_norm": 1213.98291015625,
      "learning_rate": 2.7919191919191922e-05,
      "loss": 176.116,
      "step": 6910
    },
    {
      "epoch": 0.05591512536461995,
      "grad_norm": 1919.146484375,
      "learning_rate": 2.7959595959595964e-05,
      "loss": 205.5731,
      "step": 6920
    },
    {
      "epoch": 0.05599592756890408,
      "grad_norm": 1582.1240234375,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 204.0053,
      "step": 6930
    },
    {
      "epoch": 0.056076729773188214,
      "grad_norm": 1082.2257080078125,
      "learning_rate": 2.804040404040404e-05,
      "loss": 216.7668,
      "step": 6940
    },
    {
      "epoch": 0.05615753197747234,
      "grad_norm": 1451.9715576171875,
      "learning_rate": 2.808080808080808e-05,
      "loss": 128.3858,
      "step": 6950
    },
    {
      "epoch": 0.05623833418175648,
      "grad_norm": 2580.067138671875,
      "learning_rate": 2.812121212121212e-05,
      "loss": 187.2646,
      "step": 6960
    },
    {
      "epoch": 0.05631913638604061,
      "grad_norm": 1153.5308837890625,
      "learning_rate": 2.8161616161616162e-05,
      "loss": 170.1935,
      "step": 6970
    },
    {
      "epoch": 0.056399938590324744,
      "grad_norm": 842.653076171875,
      "learning_rate": 2.82020202020202e-05,
      "loss": 229.306,
      "step": 6980
    },
    {
      "epoch": 0.05648074079460888,
      "grad_norm": 1086.96337890625,
      "learning_rate": 2.8242424242424243e-05,
      "loss": 180.9517,
      "step": 6990
    },
    {
      "epoch": 0.05656154299889301,
      "grad_norm": 963.1438598632812,
      "learning_rate": 2.8282828282828282e-05,
      "loss": 186.2078,
      "step": 7000
    },
    {
      "epoch": 0.056642345203177145,
      "grad_norm": 1010.3299560546875,
      "learning_rate": 2.8323232323232324e-05,
      "loss": 223.6001,
      "step": 7010
    },
    {
      "epoch": 0.056723147407461275,
      "grad_norm": 1217.844482421875,
      "learning_rate": 2.8363636363636363e-05,
      "loss": 179.5198,
      "step": 7020
    },
    {
      "epoch": 0.05680394961174541,
      "grad_norm": 1364.8577880859375,
      "learning_rate": 2.8404040404040405e-05,
      "loss": 212.5286,
      "step": 7030
    },
    {
      "epoch": 0.05688475181602954,
      "grad_norm": 804.541748046875,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 201.4965,
      "step": 7040
    },
    {
      "epoch": 0.056965554020313676,
      "grad_norm": 2093.808349609375,
      "learning_rate": 2.8484848484848486e-05,
      "loss": 202.5039,
      "step": 7050
    },
    {
      "epoch": 0.057046356224597805,
      "grad_norm": 1088.9471435546875,
      "learning_rate": 2.852525252525253e-05,
      "loss": 164.6322,
      "step": 7060
    },
    {
      "epoch": 0.05712715842888194,
      "grad_norm": 1510.014404296875,
      "learning_rate": 2.8565656565656567e-05,
      "loss": 246.0487,
      "step": 7070
    },
    {
      "epoch": 0.05720796063316607,
      "grad_norm": 617.3926391601562,
      "learning_rate": 2.860606060606061e-05,
      "loss": 166.5255,
      "step": 7080
    },
    {
      "epoch": 0.057288762837450206,
      "grad_norm": 1088.094482421875,
      "learning_rate": 2.864646464646465e-05,
      "loss": 180.0012,
      "step": 7090
    },
    {
      "epoch": 0.057369565041734336,
      "grad_norm": 754.35400390625,
      "learning_rate": 2.868686868686869e-05,
      "loss": 165.6718,
      "step": 7100
    },
    {
      "epoch": 0.05745036724601847,
      "grad_norm": 847.3502197265625,
      "learning_rate": 2.872727272727273e-05,
      "loss": 150.3254,
      "step": 7110
    },
    {
      "epoch": 0.0575311694503026,
      "grad_norm": 3462.79541015625,
      "learning_rate": 2.876767676767677e-05,
      "loss": 206.9913,
      "step": 7120
    },
    {
      "epoch": 0.05761197165458674,
      "grad_norm": 1302.846923828125,
      "learning_rate": 2.880808080808081e-05,
      "loss": 218.2749,
      "step": 7130
    },
    {
      "epoch": 0.05769277385887087,
      "grad_norm": 1508.3194580078125,
      "learning_rate": 2.8848484848484853e-05,
      "loss": 198.5009,
      "step": 7140
    },
    {
      "epoch": 0.057773576063155,
      "grad_norm": 1260.8990478515625,
      "learning_rate": 2.8888888888888888e-05,
      "loss": 287.7319,
      "step": 7150
    },
    {
      "epoch": 0.05785437826743914,
      "grad_norm": 2510.641357421875,
      "learning_rate": 2.8929292929292927e-05,
      "loss": 212.435,
      "step": 7160
    },
    {
      "epoch": 0.05793518047172327,
      "grad_norm": 1610.3782958984375,
      "learning_rate": 2.896969696969697e-05,
      "loss": 195.9904,
      "step": 7170
    },
    {
      "epoch": 0.058015982676007403,
      "grad_norm": 2051.1611328125,
      "learning_rate": 2.9010101010101008e-05,
      "loss": 230.746,
      "step": 7180
    },
    {
      "epoch": 0.05809678488029153,
      "grad_norm": 1708.345703125,
      "learning_rate": 2.905050505050505e-05,
      "loss": 202.1169,
      "step": 7190
    },
    {
      "epoch": 0.05817758708457567,
      "grad_norm": 991.0370483398438,
      "learning_rate": 2.909090909090909e-05,
      "loss": 182.8259,
      "step": 7200
    },
    {
      "epoch": 0.0582583892888598,
      "grad_norm": 1151.1380615234375,
      "learning_rate": 2.913131313131313e-05,
      "loss": 241.0473,
      "step": 7210
    },
    {
      "epoch": 0.058339191493143934,
      "grad_norm": 1103.3897705078125,
      "learning_rate": 2.9171717171717174e-05,
      "loss": 151.7667,
      "step": 7220
    },
    {
      "epoch": 0.05841999369742806,
      "grad_norm": 1151.0849609375,
      "learning_rate": 2.9212121212121213e-05,
      "loss": 197.9749,
      "step": 7230
    },
    {
      "epoch": 0.0585007959017122,
      "grad_norm": 983.3527221679688,
      "learning_rate": 2.9252525252525255e-05,
      "loss": 186.8989,
      "step": 7240
    },
    {
      "epoch": 0.05858159810599633,
      "grad_norm": 669.5452880859375,
      "learning_rate": 2.9292929292929294e-05,
      "loss": 179.281,
      "step": 7250
    },
    {
      "epoch": 0.058662400310280464,
      "grad_norm": 1186.9957275390625,
      "learning_rate": 2.9333333333333336e-05,
      "loss": 170.5622,
      "step": 7260
    },
    {
      "epoch": 0.0587432025145646,
      "grad_norm": 1314.4376220703125,
      "learning_rate": 2.9373737373737375e-05,
      "loss": 175.4961,
      "step": 7270
    },
    {
      "epoch": 0.05882400471884873,
      "grad_norm": 1278.834716796875,
      "learning_rate": 2.9414141414141417e-05,
      "loss": 183.9097,
      "step": 7280
    },
    {
      "epoch": 0.058904806923132866,
      "grad_norm": 1116.2734375,
      "learning_rate": 2.9454545454545456e-05,
      "loss": 143.516,
      "step": 7290
    },
    {
      "epoch": 0.058985609127416995,
      "grad_norm": 1352.628173828125,
      "learning_rate": 2.9494949494949498e-05,
      "loss": 204.3025,
      "step": 7300
    },
    {
      "epoch": 0.05906641133170113,
      "grad_norm": 1091.3201904296875,
      "learning_rate": 2.9535353535353537e-05,
      "loss": 181.4761,
      "step": 7310
    },
    {
      "epoch": 0.05914721353598526,
      "grad_norm": 1040.334716796875,
      "learning_rate": 2.957575757575758e-05,
      "loss": 170.6319,
      "step": 7320
    },
    {
      "epoch": 0.059228015740269396,
      "grad_norm": 1476.125732421875,
      "learning_rate": 2.9616161616161618e-05,
      "loss": 161.3749,
      "step": 7330
    },
    {
      "epoch": 0.059308817944553525,
      "grad_norm": 1488.0325927734375,
      "learning_rate": 2.965656565656566e-05,
      "loss": 183.3941,
      "step": 7340
    },
    {
      "epoch": 0.05938962014883766,
      "grad_norm": 481.60833740234375,
      "learning_rate": 2.96969696969697e-05,
      "loss": 199.2278,
      "step": 7350
    },
    {
      "epoch": 0.05947042235312179,
      "grad_norm": 1610.34521484375,
      "learning_rate": 2.973737373737374e-05,
      "loss": 201.723,
      "step": 7360
    },
    {
      "epoch": 0.05955122455740593,
      "grad_norm": 1576.0423583984375,
      "learning_rate": 2.9777777777777777e-05,
      "loss": 222.0852,
      "step": 7370
    },
    {
      "epoch": 0.059632026761690056,
      "grad_norm": 889.7515258789062,
      "learning_rate": 2.9818181818181816e-05,
      "loss": 193.5616,
      "step": 7380
    },
    {
      "epoch": 0.05971282896597419,
      "grad_norm": 746.6514282226562,
      "learning_rate": 2.9858585858585858e-05,
      "loss": 166.2696,
      "step": 7390
    },
    {
      "epoch": 0.05979363117025832,
      "grad_norm": 1730.69580078125,
      "learning_rate": 2.98989898989899e-05,
      "loss": 209.8799,
      "step": 7400
    },
    {
      "epoch": 0.05987443337454246,
      "grad_norm": 690.6642456054688,
      "learning_rate": 2.993939393939394e-05,
      "loss": 230.9101,
      "step": 7410
    },
    {
      "epoch": 0.05995523557882659,
      "grad_norm": 863.1697387695312,
      "learning_rate": 2.997979797979798e-05,
      "loss": 150.7177,
      "step": 7420
    },
    {
      "epoch": 0.06003603778311072,
      "grad_norm": 1267.2069091796875,
      "learning_rate": 3.002020202020202e-05,
      "loss": 210.8308,
      "step": 7430
    },
    {
      "epoch": 0.06011683998739486,
      "grad_norm": 1010.417724609375,
      "learning_rate": 3.0060606060606062e-05,
      "loss": 191.3645,
      "step": 7440
    },
    {
      "epoch": 0.06019764219167899,
      "grad_norm": 689.7382202148438,
      "learning_rate": 3.01010101010101e-05,
      "loss": 187.7134,
      "step": 7450
    },
    {
      "epoch": 0.060278444395963124,
      "grad_norm": 1864.760986328125,
      "learning_rate": 3.0141414141414144e-05,
      "loss": 214.7331,
      "step": 7460
    },
    {
      "epoch": 0.06035924660024725,
      "grad_norm": 1038.37353515625,
      "learning_rate": 3.0181818181818182e-05,
      "loss": 217.9106,
      "step": 7470
    },
    {
      "epoch": 0.06044004880453139,
      "grad_norm": 622.6604614257812,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 155.4263,
      "step": 7480
    },
    {
      "epoch": 0.06052085100881552,
      "grad_norm": 878.7538452148438,
      "learning_rate": 3.0262626262626263e-05,
      "loss": 231.1667,
      "step": 7490
    },
    {
      "epoch": 0.060601653213099654,
      "grad_norm": 1581.2225341796875,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 163.4888,
      "step": 7500
    },
    {
      "epoch": 0.060682455417383784,
      "grad_norm": 1152.7149658203125,
      "learning_rate": 3.0343434343434345e-05,
      "loss": 182.3645,
      "step": 7510
    },
    {
      "epoch": 0.06076325762166792,
      "grad_norm": 1109.6708984375,
      "learning_rate": 3.0383838383838387e-05,
      "loss": 175.0838,
      "step": 7520
    },
    {
      "epoch": 0.06084405982595205,
      "grad_norm": 1053.8270263671875,
      "learning_rate": 3.0424242424242426e-05,
      "loss": 181.691,
      "step": 7530
    },
    {
      "epoch": 0.060924862030236185,
      "grad_norm": 2113.046875,
      "learning_rate": 3.0464646464646468e-05,
      "loss": 224.7368,
      "step": 7540
    },
    {
      "epoch": 0.06100566423452032,
      "grad_norm": 1166.90478515625,
      "learning_rate": 3.050505050505051e-05,
      "loss": 206.4759,
      "step": 7550
    },
    {
      "epoch": 0.06108646643880445,
      "grad_norm": 1273.3836669921875,
      "learning_rate": 3.054545454545455e-05,
      "loss": 171.2801,
      "step": 7560
    },
    {
      "epoch": 0.061167268643088586,
      "grad_norm": 2534.885498046875,
      "learning_rate": 3.058585858585859e-05,
      "loss": 159.7586,
      "step": 7570
    },
    {
      "epoch": 0.061248070847372715,
      "grad_norm": 3763.103515625,
      "learning_rate": 3.062626262626262e-05,
      "loss": 323.4677,
      "step": 7580
    },
    {
      "epoch": 0.06132887305165685,
      "grad_norm": 1977.9522705078125,
      "learning_rate": 3.066666666666667e-05,
      "loss": 227.4736,
      "step": 7590
    },
    {
      "epoch": 0.06140967525594098,
      "grad_norm": 1690.8280029296875,
      "learning_rate": 3.070707070707071e-05,
      "loss": 192.4362,
      "step": 7600
    },
    {
      "epoch": 0.06149047746022512,
      "grad_norm": 1523.7828369140625,
      "learning_rate": 3.074747474747475e-05,
      "loss": 234.8535,
      "step": 7610
    },
    {
      "epoch": 0.061571279664509246,
      "grad_norm": 1146.36865234375,
      "learning_rate": 3.0787878787878786e-05,
      "loss": 142.9756,
      "step": 7620
    },
    {
      "epoch": 0.06165208186879338,
      "grad_norm": 895.3403930664062,
      "learning_rate": 3.082828282828283e-05,
      "loss": 201.5379,
      "step": 7630
    },
    {
      "epoch": 0.06173288407307751,
      "grad_norm": 1039.900634765625,
      "learning_rate": 3.086868686868687e-05,
      "loss": 230.2974,
      "step": 7640
    },
    {
      "epoch": 0.06181368627736165,
      "grad_norm": 1130.9986572265625,
      "learning_rate": 3.090909090909091e-05,
      "loss": 189.1531,
      "step": 7650
    },
    {
      "epoch": 0.061894488481645776,
      "grad_norm": 1224.142822265625,
      "learning_rate": 3.094949494949495e-05,
      "loss": 204.0206,
      "step": 7660
    },
    {
      "epoch": 0.06197529068592991,
      "grad_norm": 2115.472412109375,
      "learning_rate": 3.098989898989899e-05,
      "loss": 180.536,
      "step": 7670
    },
    {
      "epoch": 0.06205609289021405,
      "grad_norm": 779.9313354492188,
      "learning_rate": 3.103030303030303e-05,
      "loss": 158.623,
      "step": 7680
    },
    {
      "epoch": 0.06213689509449818,
      "grad_norm": 1337.7568359375,
      "learning_rate": 3.107070707070707e-05,
      "loss": 159.0383,
      "step": 7690
    },
    {
      "epoch": 0.062217697298782314,
      "grad_norm": 1851.648193359375,
      "learning_rate": 3.111111111111111e-05,
      "loss": 179.8161,
      "step": 7700
    },
    {
      "epoch": 0.06229849950306644,
      "grad_norm": 1469.6453857421875,
      "learning_rate": 3.1151515151515156e-05,
      "loss": 187.596,
      "step": 7710
    },
    {
      "epoch": 0.06237930170735058,
      "grad_norm": 1624.6527099609375,
      "learning_rate": 3.1191919191919194e-05,
      "loss": 214.8479,
      "step": 7720
    },
    {
      "epoch": 0.06246010391163471,
      "grad_norm": 1006.6346435546875,
      "learning_rate": 3.123232323232323e-05,
      "loss": 154.5748,
      "step": 7730
    },
    {
      "epoch": 0.06254090611591884,
      "grad_norm": 1002.5286254882812,
      "learning_rate": 3.127272727272728e-05,
      "loss": 184.4432,
      "step": 7740
    },
    {
      "epoch": 0.06262170832020297,
      "grad_norm": 1352.4193115234375,
      "learning_rate": 3.131313131313132e-05,
      "loss": 237.0036,
      "step": 7750
    },
    {
      "epoch": 0.06270251052448711,
      "grad_norm": 1084.147216796875,
      "learning_rate": 3.1353535353535357e-05,
      "loss": 164.8318,
      "step": 7760
    },
    {
      "epoch": 0.06278331272877125,
      "grad_norm": 1302.1048583984375,
      "learning_rate": 3.1393939393939395e-05,
      "loss": 164.3788,
      "step": 7770
    },
    {
      "epoch": 0.06286411493305537,
      "grad_norm": 1383.396484375,
      "learning_rate": 3.143434343434344e-05,
      "loss": 175.5805,
      "step": 7780
    },
    {
      "epoch": 0.0629449171373395,
      "grad_norm": 1246.53857421875,
      "learning_rate": 3.147474747474747e-05,
      "loss": 210.3966,
      "step": 7790
    },
    {
      "epoch": 0.06302571934162364,
      "grad_norm": 1285.145263671875,
      "learning_rate": 3.151515151515151e-05,
      "loss": 246.1903,
      "step": 7800
    },
    {
      "epoch": 0.06310652154590778,
      "grad_norm": 1620.3326416015625,
      "learning_rate": 3.155555555555556e-05,
      "loss": 196.0127,
      "step": 7810
    },
    {
      "epoch": 0.06318732375019191,
      "grad_norm": 1016.9979858398438,
      "learning_rate": 3.1595959595959596e-05,
      "loss": 210.5301,
      "step": 7820
    },
    {
      "epoch": 0.06326812595447603,
      "grad_norm": 1945.8780517578125,
      "learning_rate": 3.1636363636363635e-05,
      "loss": 239.3266,
      "step": 7830
    },
    {
      "epoch": 0.06334892815876017,
      "grad_norm": 1864.5794677734375,
      "learning_rate": 3.1676767676767674e-05,
      "loss": 193.7567,
      "step": 7840
    },
    {
      "epoch": 0.0634297303630443,
      "grad_norm": 1095.450927734375,
      "learning_rate": 3.171717171717172e-05,
      "loss": 191.1735,
      "step": 7850
    },
    {
      "epoch": 0.06351053256732844,
      "grad_norm": 1031.504150390625,
      "learning_rate": 3.175757575757576e-05,
      "loss": 185.8655,
      "step": 7860
    },
    {
      "epoch": 0.06359133477161256,
      "grad_norm": 1385.5076904296875,
      "learning_rate": 3.17979797979798e-05,
      "loss": 177.6908,
      "step": 7870
    },
    {
      "epoch": 0.0636721369758967,
      "grad_norm": 1074.5181884765625,
      "learning_rate": 3.1838383838383836e-05,
      "loss": 204.0031,
      "step": 7880
    },
    {
      "epoch": 0.06375293918018084,
      "grad_norm": 953.3314208984375,
      "learning_rate": 3.187878787878788e-05,
      "loss": 180.9185,
      "step": 7890
    },
    {
      "epoch": 0.06383374138446497,
      "grad_norm": 868.8043823242188,
      "learning_rate": 3.191919191919192e-05,
      "loss": 220.1422,
      "step": 7900
    },
    {
      "epoch": 0.0639145435887491,
      "grad_norm": 5921.494140625,
      "learning_rate": 3.195959595959596e-05,
      "loss": 167.039,
      "step": 7910
    },
    {
      "epoch": 0.06399534579303323,
      "grad_norm": 1500.1710205078125,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 138.7559,
      "step": 7920
    },
    {
      "epoch": 0.06407614799731737,
      "grad_norm": 1143.7266845703125,
      "learning_rate": 3.2040404040404044e-05,
      "loss": 195.0978,
      "step": 7930
    },
    {
      "epoch": 0.0641569502016015,
      "grad_norm": 523.0445556640625,
      "learning_rate": 3.208080808080808e-05,
      "loss": 151.1692,
      "step": 7940
    },
    {
      "epoch": 0.06423775240588563,
      "grad_norm": 2158.39013671875,
      "learning_rate": 3.212121212121212e-05,
      "loss": 236.7984,
      "step": 7950
    },
    {
      "epoch": 0.06431855461016976,
      "grad_norm": 659.3209228515625,
      "learning_rate": 3.216161616161617e-05,
      "loss": 181.1136,
      "step": 7960
    },
    {
      "epoch": 0.0643993568144539,
      "grad_norm": 608.638671875,
      "learning_rate": 3.2202020202020206e-05,
      "loss": 194.2183,
      "step": 7970
    },
    {
      "epoch": 0.06448015901873803,
      "grad_norm": 1122.7078857421875,
      "learning_rate": 3.2242424242424245e-05,
      "loss": 160.3627,
      "step": 7980
    },
    {
      "epoch": 0.06456096122302217,
      "grad_norm": 1686.80810546875,
      "learning_rate": 3.2282828282828284e-05,
      "loss": 223.456,
      "step": 7990
    },
    {
      "epoch": 0.06464176342730629,
      "grad_norm": 1573.1317138671875,
      "learning_rate": 3.232323232323233e-05,
      "loss": 224.0322,
      "step": 8000
    },
    {
      "epoch": 0.06472256563159043,
      "grad_norm": 1321.1458740234375,
      "learning_rate": 3.236363636363636e-05,
      "loss": 252.9104,
      "step": 8010
    },
    {
      "epoch": 0.06480336783587456,
      "grad_norm": 1179.701171875,
      "learning_rate": 3.24040404040404e-05,
      "loss": 223.4346,
      "step": 8020
    },
    {
      "epoch": 0.0648841700401587,
      "grad_norm": 977.9105224609375,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 152.0468,
      "step": 8030
    },
    {
      "epoch": 0.06496497224444282,
      "grad_norm": 2066.90380859375,
      "learning_rate": 3.2484848484848485e-05,
      "loss": 190.8667,
      "step": 8040
    },
    {
      "epoch": 0.06504577444872696,
      "grad_norm": 3095.08935546875,
      "learning_rate": 3.2525252525252524e-05,
      "loss": 192.0303,
      "step": 8050
    },
    {
      "epoch": 0.0651265766530111,
      "grad_norm": 2343.95947265625,
      "learning_rate": 3.256565656565656e-05,
      "loss": 157.5384,
      "step": 8060
    },
    {
      "epoch": 0.06520737885729523,
      "grad_norm": 1510.8023681640625,
      "learning_rate": 3.260606060606061e-05,
      "loss": 239.6892,
      "step": 8070
    },
    {
      "epoch": 0.06528818106157935,
      "grad_norm": 1445.597900390625,
      "learning_rate": 3.264646464646465e-05,
      "loss": 177.4803,
      "step": 8080
    },
    {
      "epoch": 0.06536898326586349,
      "grad_norm": 1667.5521240234375,
      "learning_rate": 3.2686868686868686e-05,
      "loss": 190.0459,
      "step": 8090
    },
    {
      "epoch": 0.06544978547014763,
      "grad_norm": 925.2418212890625,
      "learning_rate": 3.272727272727273e-05,
      "loss": 190.8257,
      "step": 8100
    },
    {
      "epoch": 0.06553058767443176,
      "grad_norm": 1247.4376220703125,
      "learning_rate": 3.276767676767677e-05,
      "loss": 203.9773,
      "step": 8110
    },
    {
      "epoch": 0.0656113898787159,
      "grad_norm": 1212.892822265625,
      "learning_rate": 3.280808080808081e-05,
      "loss": 203.7381,
      "step": 8120
    },
    {
      "epoch": 0.06569219208300002,
      "grad_norm": 1091.890380859375,
      "learning_rate": 3.284848484848485e-05,
      "loss": 194.8187,
      "step": 8130
    },
    {
      "epoch": 0.06577299428728416,
      "grad_norm": 2029.2864990234375,
      "learning_rate": 3.2888888888888894e-05,
      "loss": 246.8937,
      "step": 8140
    },
    {
      "epoch": 0.06585379649156829,
      "grad_norm": 920.1378784179688,
      "learning_rate": 3.292929292929293e-05,
      "loss": 215.9934,
      "step": 8150
    },
    {
      "epoch": 0.06593459869585243,
      "grad_norm": 1521.0574951171875,
      "learning_rate": 3.296969696969697e-05,
      "loss": 167.3099,
      "step": 8160
    },
    {
      "epoch": 0.06601540090013655,
      "grad_norm": 1420.7525634765625,
      "learning_rate": 3.301010101010101e-05,
      "loss": 206.7512,
      "step": 8170
    },
    {
      "epoch": 0.06609620310442069,
      "grad_norm": 840.5839233398438,
      "learning_rate": 3.3050505050505056e-05,
      "loss": 202.7185,
      "step": 8180
    },
    {
      "epoch": 0.06617700530870482,
      "grad_norm": 1193.502197265625,
      "learning_rate": 3.3090909090909095e-05,
      "loss": 160.3612,
      "step": 8190
    },
    {
      "epoch": 0.06625780751298896,
      "grad_norm": 2222.778564453125,
      "learning_rate": 3.3131313131313134e-05,
      "loss": 150.92,
      "step": 8200
    },
    {
      "epoch": 0.06633860971727308,
      "grad_norm": 776.4454956054688,
      "learning_rate": 3.317171717171717e-05,
      "loss": 159.6749,
      "step": 8210
    },
    {
      "epoch": 0.06641941192155722,
      "grad_norm": 1179.86279296875,
      "learning_rate": 3.321212121212121e-05,
      "loss": 147.1537,
      "step": 8220
    },
    {
      "epoch": 0.06650021412584135,
      "grad_norm": 1168.2757568359375,
      "learning_rate": 3.325252525252525e-05,
      "loss": 163.5715,
      "step": 8230
    },
    {
      "epoch": 0.06658101633012549,
      "grad_norm": 996.3876953125,
      "learning_rate": 3.329292929292929e-05,
      "loss": 156.5557,
      "step": 8240
    },
    {
      "epoch": 0.06666181853440963,
      "grad_norm": 1006.9996337890625,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 176.6802,
      "step": 8250
    },
    {
      "epoch": 0.06674262073869375,
      "grad_norm": 877.4000854492188,
      "learning_rate": 3.3373737373737374e-05,
      "loss": 182.8363,
      "step": 8260
    },
    {
      "epoch": 0.06682342294297788,
      "grad_norm": 2153.091552734375,
      "learning_rate": 3.341414141414141e-05,
      "loss": 184.7595,
      "step": 8270
    },
    {
      "epoch": 0.06690422514726202,
      "grad_norm": 1884.7989501953125,
      "learning_rate": 3.345454545454546e-05,
      "loss": 197.672,
      "step": 8280
    },
    {
      "epoch": 0.06698502735154616,
      "grad_norm": 1494.185791015625,
      "learning_rate": 3.34949494949495e-05,
      "loss": 178.7865,
      "step": 8290
    },
    {
      "epoch": 0.06706582955583028,
      "grad_norm": 2600.398193359375,
      "learning_rate": 3.3535353535353536e-05,
      "loss": 222.7885,
      "step": 8300
    },
    {
      "epoch": 0.06714663176011441,
      "grad_norm": 1300.013671875,
      "learning_rate": 3.3575757575757575e-05,
      "loss": 182.5449,
      "step": 8310
    },
    {
      "epoch": 0.06722743396439855,
      "grad_norm": 2145.218505859375,
      "learning_rate": 3.361616161616162e-05,
      "loss": 281.7514,
      "step": 8320
    },
    {
      "epoch": 0.06730823616868269,
      "grad_norm": 1519.411865234375,
      "learning_rate": 3.365656565656566e-05,
      "loss": 167.7319,
      "step": 8330
    },
    {
      "epoch": 0.06738903837296681,
      "grad_norm": 750.0274047851562,
      "learning_rate": 3.36969696969697e-05,
      "loss": 173.5281,
      "step": 8340
    },
    {
      "epoch": 0.06746984057725094,
      "grad_norm": 1222.1435546875,
      "learning_rate": 3.373737373737374e-05,
      "loss": 162.324,
      "step": 8350
    },
    {
      "epoch": 0.06755064278153508,
      "grad_norm": 955.3302001953125,
      "learning_rate": 3.377777777777778e-05,
      "loss": 184.845,
      "step": 8360
    },
    {
      "epoch": 0.06763144498581922,
      "grad_norm": 1112.2943115234375,
      "learning_rate": 3.381818181818182e-05,
      "loss": 160.8671,
      "step": 8370
    },
    {
      "epoch": 0.06771224719010335,
      "grad_norm": 1163.462646484375,
      "learning_rate": 3.385858585858586e-05,
      "loss": 147.4451,
      "step": 8380
    },
    {
      "epoch": 0.06779304939438748,
      "grad_norm": 925.3172607421875,
      "learning_rate": 3.38989898989899e-05,
      "loss": 175.2849,
      "step": 8390
    },
    {
      "epoch": 0.06787385159867161,
      "grad_norm": 1990.568359375,
      "learning_rate": 3.3939393939393945e-05,
      "loss": 199.808,
      "step": 8400
    },
    {
      "epoch": 0.06795465380295575,
      "grad_norm": 2128.471923828125,
      "learning_rate": 3.3979797979797984e-05,
      "loss": 205.3155,
      "step": 8410
    },
    {
      "epoch": 0.06803545600723988,
      "grad_norm": 2691.37353515625,
      "learning_rate": 3.402020202020202e-05,
      "loss": 216.1895,
      "step": 8420
    },
    {
      "epoch": 0.068116258211524,
      "grad_norm": 3363.869140625,
      "learning_rate": 3.406060606060606e-05,
      "loss": 201.8374,
      "step": 8430
    },
    {
      "epoch": 0.06819706041580814,
      "grad_norm": 1438.0633544921875,
      "learning_rate": 3.41010101010101e-05,
      "loss": 222.5396,
      "step": 8440
    },
    {
      "epoch": 0.06827786262009228,
      "grad_norm": 1703.8653564453125,
      "learning_rate": 3.414141414141414e-05,
      "loss": 213.8756,
      "step": 8450
    },
    {
      "epoch": 0.06835866482437641,
      "grad_norm": 1938.7177734375,
      "learning_rate": 3.4181818181818185e-05,
      "loss": 215.6468,
      "step": 8460
    },
    {
      "epoch": 0.06843946702866054,
      "grad_norm": 851.2493896484375,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 182.1204,
      "step": 8470
    },
    {
      "epoch": 0.06852026923294467,
      "grad_norm": 1202.3365478515625,
      "learning_rate": 3.426262626262626e-05,
      "loss": 183.2559,
      "step": 8480
    },
    {
      "epoch": 0.06860107143722881,
      "grad_norm": 1543.7257080078125,
      "learning_rate": 3.43030303030303e-05,
      "loss": 239.1145,
      "step": 8490
    },
    {
      "epoch": 0.06868187364151294,
      "grad_norm": 748.701171875,
      "learning_rate": 3.434343434343435e-05,
      "loss": 285.6954,
      "step": 8500
    },
    {
      "epoch": 0.06876267584579707,
      "grad_norm": 5747.30224609375,
      "learning_rate": 3.4383838383838386e-05,
      "loss": 194.7742,
      "step": 8510
    },
    {
      "epoch": 0.0688434780500812,
      "grad_norm": 1587.478271484375,
      "learning_rate": 3.4424242424242425e-05,
      "loss": 213.0056,
      "step": 8520
    },
    {
      "epoch": 0.06892428025436534,
      "grad_norm": 907.9869995117188,
      "learning_rate": 3.4464646464646463e-05,
      "loss": 183.2854,
      "step": 8530
    },
    {
      "epoch": 0.06900508245864947,
      "grad_norm": 1065.7462158203125,
      "learning_rate": 3.450505050505051e-05,
      "loss": 147.3901,
      "step": 8540
    },
    {
      "epoch": 0.06908588466293361,
      "grad_norm": 1654.4375,
      "learning_rate": 3.454545454545455e-05,
      "loss": 152.4811,
      "step": 8550
    },
    {
      "epoch": 0.06916668686721773,
      "grad_norm": 1075.144775390625,
      "learning_rate": 3.458585858585859e-05,
      "loss": 217.3591,
      "step": 8560
    },
    {
      "epoch": 0.06924748907150187,
      "grad_norm": 1216.4287109375,
      "learning_rate": 3.4626262626262626e-05,
      "loss": 184.7878,
      "step": 8570
    },
    {
      "epoch": 0.069328291275786,
      "grad_norm": 1143.3253173828125,
      "learning_rate": 3.466666666666667e-05,
      "loss": 185.8244,
      "step": 8580
    },
    {
      "epoch": 0.06940909348007014,
      "grad_norm": 2943.891357421875,
      "learning_rate": 3.470707070707071e-05,
      "loss": 203.6546,
      "step": 8590
    },
    {
      "epoch": 0.06948989568435426,
      "grad_norm": 562.7566528320312,
      "learning_rate": 3.474747474747475e-05,
      "loss": 224.7354,
      "step": 8600
    },
    {
      "epoch": 0.0695706978886384,
      "grad_norm": 1995.734130859375,
      "learning_rate": 3.4787878787878795e-05,
      "loss": 178.8969,
      "step": 8610
    },
    {
      "epoch": 0.06965150009292254,
      "grad_norm": 1593.8944091796875,
      "learning_rate": 3.4828282828282834e-05,
      "loss": 164.9677,
      "step": 8620
    },
    {
      "epoch": 0.06973230229720667,
      "grad_norm": 715.6300048828125,
      "learning_rate": 3.486868686868687e-05,
      "loss": 147.6377,
      "step": 8630
    },
    {
      "epoch": 0.0698131045014908,
      "grad_norm": 3263.21044921875,
      "learning_rate": 3.490909090909091e-05,
      "loss": 235.9829,
      "step": 8640
    },
    {
      "epoch": 0.06989390670577493,
      "grad_norm": 3928.5576171875,
      "learning_rate": 3.494949494949495e-05,
      "loss": 220.889,
      "step": 8650
    },
    {
      "epoch": 0.06997470891005907,
      "grad_norm": 1176.6265869140625,
      "learning_rate": 3.498989898989899e-05,
      "loss": 173.9785,
      "step": 8660
    },
    {
      "epoch": 0.0700555111143432,
      "grad_norm": 1472.33349609375,
      "learning_rate": 3.503030303030303e-05,
      "loss": 166.185,
      "step": 8670
    },
    {
      "epoch": 0.07013631331862734,
      "grad_norm": 943.4843139648438,
      "learning_rate": 3.5070707070707073e-05,
      "loss": 196.9754,
      "step": 8680
    },
    {
      "epoch": 0.07021711552291146,
      "grad_norm": 1376.169189453125,
      "learning_rate": 3.511111111111111e-05,
      "loss": 249.9044,
      "step": 8690
    },
    {
      "epoch": 0.0702979177271956,
      "grad_norm": 862.705078125,
      "learning_rate": 3.515151515151515e-05,
      "loss": 134.6473,
      "step": 8700
    },
    {
      "epoch": 0.07037871993147973,
      "grad_norm": 1661.3258056640625,
      "learning_rate": 3.519191919191919e-05,
      "loss": 260.4335,
      "step": 8710
    },
    {
      "epoch": 0.07045952213576387,
      "grad_norm": 858.2864379882812,
      "learning_rate": 3.5232323232323236e-05,
      "loss": 156.1466,
      "step": 8720
    },
    {
      "epoch": 0.07054032434004799,
      "grad_norm": 1033.8033447265625,
      "learning_rate": 3.5272727272727274e-05,
      "loss": 158.8132,
      "step": 8730
    },
    {
      "epoch": 0.07062112654433213,
      "grad_norm": 2244.4833984375,
      "learning_rate": 3.531313131313131e-05,
      "loss": 185.2664,
      "step": 8740
    },
    {
      "epoch": 0.07070192874861626,
      "grad_norm": 828.0194091796875,
      "learning_rate": 3.535353535353535e-05,
      "loss": 189.655,
      "step": 8750
    },
    {
      "epoch": 0.0707827309529004,
      "grad_norm": 764.8339233398438,
      "learning_rate": 3.53939393939394e-05,
      "loss": 169.7833,
      "step": 8760
    },
    {
      "epoch": 0.07086353315718452,
      "grad_norm": 1434.6533203125,
      "learning_rate": 3.543434343434344e-05,
      "loss": 169.9032,
      "step": 8770
    },
    {
      "epoch": 0.07094433536146866,
      "grad_norm": 1811.5740966796875,
      "learning_rate": 3.5474747474747475e-05,
      "loss": 256.6501,
      "step": 8780
    },
    {
      "epoch": 0.0710251375657528,
      "grad_norm": 923.5958251953125,
      "learning_rate": 3.551515151515152e-05,
      "loss": 157.0084,
      "step": 8790
    },
    {
      "epoch": 0.07110593977003693,
      "grad_norm": 1671.8385009765625,
      "learning_rate": 3.555555555555556e-05,
      "loss": 206.8505,
      "step": 8800
    },
    {
      "epoch": 0.07118674197432107,
      "grad_norm": 2508.17626953125,
      "learning_rate": 3.55959595959596e-05,
      "loss": 204.866,
      "step": 8810
    },
    {
      "epoch": 0.07126754417860519,
      "grad_norm": 852.1519775390625,
      "learning_rate": 3.563636363636364e-05,
      "loss": 147.0554,
      "step": 8820
    },
    {
      "epoch": 0.07134834638288932,
      "grad_norm": 925.072021484375,
      "learning_rate": 3.567676767676768e-05,
      "loss": 205.7363,
      "step": 8830
    },
    {
      "epoch": 0.07142914858717346,
      "grad_norm": 1310.513916015625,
      "learning_rate": 3.571717171717172e-05,
      "loss": 252.9427,
      "step": 8840
    },
    {
      "epoch": 0.0715099507914576,
      "grad_norm": 1795.476806640625,
      "learning_rate": 3.575757575757576e-05,
      "loss": 184.422,
      "step": 8850
    },
    {
      "epoch": 0.07159075299574172,
      "grad_norm": 1071.3101806640625,
      "learning_rate": 3.57979797979798e-05,
      "loss": 165.8845,
      "step": 8860
    },
    {
      "epoch": 0.07167155520002585,
      "grad_norm": 724.8527221679688,
      "learning_rate": 3.583838383838384e-05,
      "loss": 180.9637,
      "step": 8870
    },
    {
      "epoch": 0.07175235740430999,
      "grad_norm": 999.9872436523438,
      "learning_rate": 3.587878787878788e-05,
      "loss": 186.5519,
      "step": 8880
    },
    {
      "epoch": 0.07183315960859413,
      "grad_norm": 1380.1075439453125,
      "learning_rate": 3.5919191919191916e-05,
      "loss": 224.0065,
      "step": 8890
    },
    {
      "epoch": 0.07191396181287825,
      "grad_norm": 1093.9498291015625,
      "learning_rate": 3.595959595959596e-05,
      "loss": 165.8288,
      "step": 8900
    },
    {
      "epoch": 0.07199476401716239,
      "grad_norm": 2711.7353515625,
      "learning_rate": 3.6e-05,
      "loss": 163.5886,
      "step": 8910
    },
    {
      "epoch": 0.07207556622144652,
      "grad_norm": 1537.253662109375,
      "learning_rate": 3.604040404040404e-05,
      "loss": 165.1541,
      "step": 8920
    },
    {
      "epoch": 0.07215636842573066,
      "grad_norm": 1100.548095703125,
      "learning_rate": 3.608080808080808e-05,
      "loss": 169.8857,
      "step": 8930
    },
    {
      "epoch": 0.0722371706300148,
      "grad_norm": 1088.0587158203125,
      "learning_rate": 3.6121212121212124e-05,
      "loss": 186.5488,
      "step": 8940
    },
    {
      "epoch": 0.07231797283429892,
      "grad_norm": 1160.9769287109375,
      "learning_rate": 3.616161616161616e-05,
      "loss": 190.4372,
      "step": 8950
    },
    {
      "epoch": 0.07239877503858305,
      "grad_norm": 1117.179443359375,
      "learning_rate": 3.62020202020202e-05,
      "loss": 134.2678,
      "step": 8960
    },
    {
      "epoch": 0.07247957724286719,
      "grad_norm": 1956.5089111328125,
      "learning_rate": 3.624242424242425e-05,
      "loss": 172.5972,
      "step": 8970
    },
    {
      "epoch": 0.07256037944715132,
      "grad_norm": 2694.300537109375,
      "learning_rate": 3.6282828282828286e-05,
      "loss": 200.7171,
      "step": 8980
    },
    {
      "epoch": 0.07264118165143545,
      "grad_norm": 1112.63720703125,
      "learning_rate": 3.6323232323232325e-05,
      "loss": 151.1802,
      "step": 8990
    },
    {
      "epoch": 0.07272198385571958,
      "grad_norm": 2176.663330078125,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 162.1838,
      "step": 9000
    },
    {
      "epoch": 0.07280278606000372,
      "grad_norm": 1126.674072265625,
      "learning_rate": 3.640404040404041e-05,
      "loss": 180.7603,
      "step": 9010
    },
    {
      "epoch": 0.07288358826428785,
      "grad_norm": 1244.1241455078125,
      "learning_rate": 3.644444444444445e-05,
      "loss": 114.9891,
      "step": 9020
    },
    {
      "epoch": 0.07296439046857198,
      "grad_norm": 2782.807373046875,
      "learning_rate": 3.648484848484849e-05,
      "loss": 199.3326,
      "step": 9030
    },
    {
      "epoch": 0.07304519267285611,
      "grad_norm": 1218.721435546875,
      "learning_rate": 3.6525252525252526e-05,
      "loss": 143.516,
      "step": 9040
    },
    {
      "epoch": 0.07312599487714025,
      "grad_norm": 799.7451782226562,
      "learning_rate": 3.656565656565657e-05,
      "loss": 131.4932,
      "step": 9050
    },
    {
      "epoch": 0.07320679708142439,
      "grad_norm": 547.2342529296875,
      "learning_rate": 3.660606060606061e-05,
      "loss": 151.0718,
      "step": 9060
    },
    {
      "epoch": 0.07328759928570851,
      "grad_norm": 1480.14501953125,
      "learning_rate": 3.664646464646464e-05,
      "loss": 201.2714,
      "step": 9070
    },
    {
      "epoch": 0.07336840148999264,
      "grad_norm": 516.29931640625,
      "learning_rate": 3.668686868686869e-05,
      "loss": 174.2223,
      "step": 9080
    },
    {
      "epoch": 0.07344920369427678,
      "grad_norm": 912.0347900390625,
      "learning_rate": 3.672727272727273e-05,
      "loss": 180.7495,
      "step": 9090
    },
    {
      "epoch": 0.07353000589856092,
      "grad_norm": 1243.236083984375,
      "learning_rate": 3.6767676767676766e-05,
      "loss": 189.5693,
      "step": 9100
    },
    {
      "epoch": 0.07361080810284505,
      "grad_norm": 742.5632934570312,
      "learning_rate": 3.6808080808080805e-05,
      "loss": 194.6184,
      "step": 9110
    },
    {
      "epoch": 0.07369161030712917,
      "grad_norm": 1145.2069091796875,
      "learning_rate": 3.684848484848485e-05,
      "loss": 185.7565,
      "step": 9120
    },
    {
      "epoch": 0.07377241251141331,
      "grad_norm": 1086.181640625,
      "learning_rate": 3.688888888888889e-05,
      "loss": 167.5214,
      "step": 9130
    },
    {
      "epoch": 0.07385321471569745,
      "grad_norm": 1321.85400390625,
      "learning_rate": 3.692929292929293e-05,
      "loss": 221.075,
      "step": 9140
    },
    {
      "epoch": 0.07393401691998158,
      "grad_norm": 2168.907958984375,
      "learning_rate": 3.6969696969696974e-05,
      "loss": 219.8519,
      "step": 9150
    },
    {
      "epoch": 0.0740148191242657,
      "grad_norm": 1007.1217041015625,
      "learning_rate": 3.701010101010101e-05,
      "loss": 194.1429,
      "step": 9160
    },
    {
      "epoch": 0.07409562132854984,
      "grad_norm": 1099.997802734375,
      "learning_rate": 3.705050505050505e-05,
      "loss": 220.093,
      "step": 9170
    },
    {
      "epoch": 0.07417642353283398,
      "grad_norm": 745.4526977539062,
      "learning_rate": 3.709090909090909e-05,
      "loss": 143.6606,
      "step": 9180
    },
    {
      "epoch": 0.07425722573711811,
      "grad_norm": 2050.18212890625,
      "learning_rate": 3.7131313131313136e-05,
      "loss": 222.1462,
      "step": 9190
    },
    {
      "epoch": 0.07433802794140223,
      "grad_norm": 2366.43896484375,
      "learning_rate": 3.7171717171717175e-05,
      "loss": 191.5783,
      "step": 9200
    },
    {
      "epoch": 0.07441883014568637,
      "grad_norm": 905.9826049804688,
      "learning_rate": 3.7212121212121214e-05,
      "loss": 168.3594,
      "step": 9210
    },
    {
      "epoch": 0.07449963234997051,
      "grad_norm": 1320.197509765625,
      "learning_rate": 3.725252525252525e-05,
      "loss": 140.7596,
      "step": 9220
    },
    {
      "epoch": 0.07458043455425464,
      "grad_norm": 3290.6748046875,
      "learning_rate": 3.72929292929293e-05,
      "loss": 161.625,
      "step": 9230
    },
    {
      "epoch": 0.07466123675853878,
      "grad_norm": 1262.2984619140625,
      "learning_rate": 3.733333333333334e-05,
      "loss": 163.4713,
      "step": 9240
    },
    {
      "epoch": 0.0747420389628229,
      "grad_norm": 1434.2923583984375,
      "learning_rate": 3.7373737373737376e-05,
      "loss": 168.6003,
      "step": 9250
    },
    {
      "epoch": 0.07482284116710704,
      "grad_norm": 716.86083984375,
      "learning_rate": 3.7414141414141415e-05,
      "loss": 165.9515,
      "step": 9260
    },
    {
      "epoch": 0.07490364337139117,
      "grad_norm": 4190.2119140625,
      "learning_rate": 3.745454545454546e-05,
      "loss": 201.2678,
      "step": 9270
    },
    {
      "epoch": 0.07498444557567531,
      "grad_norm": 1093.3636474609375,
      "learning_rate": 3.74949494949495e-05,
      "loss": 145.2475,
      "step": 9280
    },
    {
      "epoch": 0.07506524777995943,
      "grad_norm": 1217.2969970703125,
      "learning_rate": 3.753535353535353e-05,
      "loss": 219.3316,
      "step": 9290
    },
    {
      "epoch": 0.07514604998424357,
      "grad_norm": 1122.0589599609375,
      "learning_rate": 3.757575757575758e-05,
      "loss": 201.7304,
      "step": 9300
    },
    {
      "epoch": 0.0752268521885277,
      "grad_norm": 779.33447265625,
      "learning_rate": 3.7616161616161616e-05,
      "loss": 199.0558,
      "step": 9310
    },
    {
      "epoch": 0.07530765439281184,
      "grad_norm": 1110.2554931640625,
      "learning_rate": 3.7656565656565655e-05,
      "loss": 190.4773,
      "step": 9320
    },
    {
      "epoch": 0.07538845659709596,
      "grad_norm": 1762.0330810546875,
      "learning_rate": 3.76969696969697e-05,
      "loss": 151.1484,
      "step": 9330
    },
    {
      "epoch": 0.0754692588013801,
      "grad_norm": 1020.6593627929688,
      "learning_rate": 3.773737373737374e-05,
      "loss": 169.2854,
      "step": 9340
    },
    {
      "epoch": 0.07555006100566423,
      "grad_norm": 1202.7464599609375,
      "learning_rate": 3.777777777777778e-05,
      "loss": 164.9156,
      "step": 9350
    },
    {
      "epoch": 0.07563086320994837,
      "grad_norm": 1249.380126953125,
      "learning_rate": 3.781818181818182e-05,
      "loss": 227.8868,
      "step": 9360
    },
    {
      "epoch": 0.0757116654142325,
      "grad_norm": 1847.7193603515625,
      "learning_rate": 3.785858585858586e-05,
      "loss": 176.2305,
      "step": 9370
    },
    {
      "epoch": 0.07579246761851663,
      "grad_norm": 1446.559814453125,
      "learning_rate": 3.78989898989899e-05,
      "loss": 147.3983,
      "step": 9380
    },
    {
      "epoch": 0.07587326982280077,
      "grad_norm": 1216.9862060546875,
      "learning_rate": 3.793939393939394e-05,
      "loss": 165.7989,
      "step": 9390
    },
    {
      "epoch": 0.0759540720270849,
      "grad_norm": 2848.1337890625,
      "learning_rate": 3.797979797979798e-05,
      "loss": 157.2326,
      "step": 9400
    },
    {
      "epoch": 0.07603487423136904,
      "grad_norm": 940.07177734375,
      "learning_rate": 3.8020202020202025e-05,
      "loss": 221.2202,
      "step": 9410
    },
    {
      "epoch": 0.07611567643565316,
      "grad_norm": 1160.6881103515625,
      "learning_rate": 3.8060606060606064e-05,
      "loss": 158.6488,
      "step": 9420
    },
    {
      "epoch": 0.0761964786399373,
      "grad_norm": 2879.822998046875,
      "learning_rate": 3.81010101010101e-05,
      "loss": 223.0843,
      "step": 9430
    },
    {
      "epoch": 0.07627728084422143,
      "grad_norm": 1498.8753662109375,
      "learning_rate": 3.814141414141414e-05,
      "loss": 161.5878,
      "step": 9440
    },
    {
      "epoch": 0.07635808304850557,
      "grad_norm": 1229.455078125,
      "learning_rate": 3.818181818181819e-05,
      "loss": 198.7026,
      "step": 9450
    },
    {
      "epoch": 0.07643888525278969,
      "grad_norm": 929.4089965820312,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 251.8946,
      "step": 9460
    },
    {
      "epoch": 0.07651968745707383,
      "grad_norm": 934.8760375976562,
      "learning_rate": 3.8262626262626265e-05,
      "loss": 162.8998,
      "step": 9470
    },
    {
      "epoch": 0.07660048966135796,
      "grad_norm": 915.323974609375,
      "learning_rate": 3.830303030303031e-05,
      "loss": 171.6096,
      "step": 9480
    },
    {
      "epoch": 0.0766812918656421,
      "grad_norm": 1115.416748046875,
      "learning_rate": 3.834343434343435e-05,
      "loss": 163.5847,
      "step": 9490
    },
    {
      "epoch": 0.07676209406992623,
      "grad_norm": 2073.390869140625,
      "learning_rate": 3.838383838383838e-05,
      "loss": 261.5975,
      "step": 9500
    },
    {
      "epoch": 0.07684289627421036,
      "grad_norm": 1260.9718017578125,
      "learning_rate": 3.842424242424243e-05,
      "loss": 131.5614,
      "step": 9510
    },
    {
      "epoch": 0.07692369847849449,
      "grad_norm": 1239.3568115234375,
      "learning_rate": 3.8464646464646466e-05,
      "loss": 164.9494,
      "step": 9520
    },
    {
      "epoch": 0.07700450068277863,
      "grad_norm": 1221.8023681640625,
      "learning_rate": 3.8505050505050505e-05,
      "loss": 245.5353,
      "step": 9530
    },
    {
      "epoch": 0.07708530288706276,
      "grad_norm": 533.4956665039062,
      "learning_rate": 3.8545454545454544e-05,
      "loss": 149.1819,
      "step": 9540
    },
    {
      "epoch": 0.07716610509134689,
      "grad_norm": 1410.545166015625,
      "learning_rate": 3.858585858585859e-05,
      "loss": 172.351,
      "step": 9550
    },
    {
      "epoch": 0.07724690729563102,
      "grad_norm": 927.8252563476562,
      "learning_rate": 3.862626262626263e-05,
      "loss": 152.7654,
      "step": 9560
    },
    {
      "epoch": 0.07732770949991516,
      "grad_norm": 1244.0257568359375,
      "learning_rate": 3.866666666666667e-05,
      "loss": 171.8292,
      "step": 9570
    },
    {
      "epoch": 0.0774085117041993,
      "grad_norm": 664.2005615234375,
      "learning_rate": 3.8707070707070706e-05,
      "loss": 109.5876,
      "step": 9580
    },
    {
      "epoch": 0.07748931390848342,
      "grad_norm": 1334.42626953125,
      "learning_rate": 3.874747474747475e-05,
      "loss": 129.1926,
      "step": 9590
    },
    {
      "epoch": 0.07757011611276755,
      "grad_norm": 1236.3963623046875,
      "learning_rate": 3.878787878787879e-05,
      "loss": 195.8564,
      "step": 9600
    },
    {
      "epoch": 0.07765091831705169,
      "grad_norm": 517.2808227539062,
      "learning_rate": 3.882828282828283e-05,
      "loss": 213.7913,
      "step": 9610
    },
    {
      "epoch": 0.07773172052133583,
      "grad_norm": 1618.37890625,
      "learning_rate": 3.886868686868687e-05,
      "loss": 187.1561,
      "step": 9620
    },
    {
      "epoch": 0.07781252272561995,
      "grad_norm": 1312.72705078125,
      "learning_rate": 3.8909090909090914e-05,
      "loss": 192.0711,
      "step": 9630
    },
    {
      "epoch": 0.07789332492990408,
      "grad_norm": 970.0208129882812,
      "learning_rate": 3.894949494949495e-05,
      "loss": 150.2422,
      "step": 9640
    },
    {
      "epoch": 0.07797412713418822,
      "grad_norm": 1302.1982421875,
      "learning_rate": 3.898989898989899e-05,
      "loss": 149.0684,
      "step": 9650
    },
    {
      "epoch": 0.07805492933847236,
      "grad_norm": 3663.638427734375,
      "learning_rate": 3.903030303030304e-05,
      "loss": 197.8055,
      "step": 9660
    },
    {
      "epoch": 0.07813573154275649,
      "grad_norm": 1510.8233642578125,
      "learning_rate": 3.9070707070707076e-05,
      "loss": 217.5386,
      "step": 9670
    },
    {
      "epoch": 0.07821653374704061,
      "grad_norm": 4103.01904296875,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 198.1855,
      "step": 9680
    },
    {
      "epoch": 0.07829733595132475,
      "grad_norm": 1483.061279296875,
      "learning_rate": 3.9151515151515153e-05,
      "loss": 200.777,
      "step": 9690
    },
    {
      "epoch": 0.07837813815560889,
      "grad_norm": 1666.9429931640625,
      "learning_rate": 3.91919191919192e-05,
      "loss": 233.7256,
      "step": 9700
    },
    {
      "epoch": 0.07845894035989302,
      "grad_norm": 1422.605224609375,
      "learning_rate": 3.923232323232323e-05,
      "loss": 198.8265,
      "step": 9710
    },
    {
      "epoch": 0.07853974256417715,
      "grad_norm": 955.90869140625,
      "learning_rate": 3.927272727272727e-05,
      "loss": 178.4951,
      "step": 9720
    },
    {
      "epoch": 0.07862054476846128,
      "grad_norm": 768.09228515625,
      "learning_rate": 3.9313131313131316e-05,
      "loss": 169.9865,
      "step": 9730
    },
    {
      "epoch": 0.07870134697274542,
      "grad_norm": 2241.572998046875,
      "learning_rate": 3.9353535353535355e-05,
      "loss": 171.3657,
      "step": 9740
    },
    {
      "epoch": 0.07878214917702955,
      "grad_norm": 828.8177490234375,
      "learning_rate": 3.939393939393939e-05,
      "loss": 212.0979,
      "step": 9750
    },
    {
      "epoch": 0.07886295138131368,
      "grad_norm": 1248.2691650390625,
      "learning_rate": 3.943434343434343e-05,
      "loss": 144.412,
      "step": 9760
    },
    {
      "epoch": 0.07894375358559781,
      "grad_norm": 1106.5013427734375,
      "learning_rate": 3.947474747474748e-05,
      "loss": 170.0778,
      "step": 9770
    },
    {
      "epoch": 0.07902455578988195,
      "grad_norm": 1183.6558837890625,
      "learning_rate": 3.951515151515152e-05,
      "loss": 183.0114,
      "step": 9780
    },
    {
      "epoch": 0.07910535799416608,
      "grad_norm": 790.3275146484375,
      "learning_rate": 3.9555555555555556e-05,
      "loss": 148.8441,
      "step": 9790
    },
    {
      "epoch": 0.07918616019845022,
      "grad_norm": 1040.2529296875,
      "learning_rate": 3.9595959595959594e-05,
      "loss": 181.9307,
      "step": 9800
    },
    {
      "epoch": 0.07926696240273434,
      "grad_norm": 1023.8417358398438,
      "learning_rate": 3.963636363636364e-05,
      "loss": 160.768,
      "step": 9810
    },
    {
      "epoch": 0.07934776460701848,
      "grad_norm": 1530.9327392578125,
      "learning_rate": 3.967676767676768e-05,
      "loss": 189.5177,
      "step": 9820
    },
    {
      "epoch": 0.07942856681130261,
      "grad_norm": 1020.0157470703125,
      "learning_rate": 3.971717171717172e-05,
      "loss": 175.2649,
      "step": 9830
    },
    {
      "epoch": 0.07950936901558675,
      "grad_norm": 1115.394287109375,
      "learning_rate": 3.975757575757576e-05,
      "loss": 180.9754,
      "step": 9840
    },
    {
      "epoch": 0.07959017121987087,
      "grad_norm": 596.3128051757812,
      "learning_rate": 3.97979797979798e-05,
      "loss": 190.0906,
      "step": 9850
    },
    {
      "epoch": 0.07967097342415501,
      "grad_norm": 1229.2056884765625,
      "learning_rate": 3.983838383838384e-05,
      "loss": 153.1082,
      "step": 9860
    },
    {
      "epoch": 0.07975177562843914,
      "grad_norm": 1460.8936767578125,
      "learning_rate": 3.987878787878788e-05,
      "loss": 176.2776,
      "step": 9870
    },
    {
      "epoch": 0.07983257783272328,
      "grad_norm": 4583.7373046875,
      "learning_rate": 3.9919191919191926e-05,
      "loss": 163.6398,
      "step": 9880
    },
    {
      "epoch": 0.0799133800370074,
      "grad_norm": 1115.4329833984375,
      "learning_rate": 3.9959595959595964e-05,
      "loss": 174.1616,
      "step": 9890
    },
    {
      "epoch": 0.07999418224129154,
      "grad_norm": 2389.8232421875,
      "learning_rate": 4e-05,
      "loss": 163.833,
      "step": 9900
    },
    {
      "epoch": 0.08007498444557568,
      "grad_norm": 2180.552978515625,
      "learning_rate": 4.004040404040404e-05,
      "loss": 189.1826,
      "step": 9910
    },
    {
      "epoch": 0.08015578664985981,
      "grad_norm": 2478.71533203125,
      "learning_rate": 4.008080808080809e-05,
      "loss": 262.7207,
      "step": 9920
    },
    {
      "epoch": 0.08023658885414395,
      "grad_norm": 952.7739868164062,
      "learning_rate": 4.012121212121212e-05,
      "loss": 152.5153,
      "step": 9930
    },
    {
      "epoch": 0.08031739105842807,
      "grad_norm": 1591.5555419921875,
      "learning_rate": 4.016161616161616e-05,
      "loss": 161.1086,
      "step": 9940
    },
    {
      "epoch": 0.0803981932627122,
      "grad_norm": 2502.85400390625,
      "learning_rate": 4.0202020202020204e-05,
      "loss": 180.6969,
      "step": 9950
    },
    {
      "epoch": 0.08047899546699634,
      "grad_norm": 1107.453857421875,
      "learning_rate": 4.024242424242424e-05,
      "loss": 245.7,
      "step": 9960
    },
    {
      "epoch": 0.08055979767128048,
      "grad_norm": 732.3422241210938,
      "learning_rate": 4.028282828282828e-05,
      "loss": 128.3301,
      "step": 9970
    },
    {
      "epoch": 0.0806405998755646,
      "grad_norm": 852.4077758789062,
      "learning_rate": 4.032323232323232e-05,
      "loss": 171.862,
      "step": 9980
    },
    {
      "epoch": 0.08072140207984874,
      "grad_norm": 974.1900634765625,
      "learning_rate": 4.0363636363636367e-05,
      "loss": 245.3339,
      "step": 9990
    },
    {
      "epoch": 0.08080220428413287,
      "grad_norm": 917.6868286132812,
      "learning_rate": 4.0404040404040405e-05,
      "loss": 151.893,
      "step": 10000
    },
    {
      "epoch": 0.08088300648841701,
      "grad_norm": 1009.7120971679688,
      "learning_rate": 4.0444444444444444e-05,
      "loss": 196.4933,
      "step": 10010
    },
    {
      "epoch": 0.08096380869270113,
      "grad_norm": 2075.980224609375,
      "learning_rate": 4.048484848484849e-05,
      "loss": 190.6988,
      "step": 10020
    },
    {
      "epoch": 0.08104461089698527,
      "grad_norm": 2236.189697265625,
      "learning_rate": 4.052525252525253e-05,
      "loss": 173.7965,
      "step": 10030
    },
    {
      "epoch": 0.0811254131012694,
      "grad_norm": 1387.155517578125,
      "learning_rate": 4.056565656565657e-05,
      "loss": 287.3005,
      "step": 10040
    },
    {
      "epoch": 0.08120621530555354,
      "grad_norm": 1775.0162353515625,
      "learning_rate": 4.0606060606060606e-05,
      "loss": 161.304,
      "step": 10050
    },
    {
      "epoch": 0.08128701750983768,
      "grad_norm": 2991.034423828125,
      "learning_rate": 4.064646464646465e-05,
      "loss": 183.5822,
      "step": 10060
    },
    {
      "epoch": 0.0813678197141218,
      "grad_norm": 2393.831298828125,
      "learning_rate": 4.068686868686869e-05,
      "loss": 216.6813,
      "step": 10070
    },
    {
      "epoch": 0.08144862191840593,
      "grad_norm": 753.2874755859375,
      "learning_rate": 4.072727272727273e-05,
      "loss": 176.5726,
      "step": 10080
    },
    {
      "epoch": 0.08152942412269007,
      "grad_norm": 917.6944580078125,
      "learning_rate": 4.076767676767677e-05,
      "loss": 182.8443,
      "step": 10090
    },
    {
      "epoch": 0.0816102263269742,
      "grad_norm": 1497.6976318359375,
      "learning_rate": 4.0808080808080814e-05,
      "loss": 179.7216,
      "step": 10100
    },
    {
      "epoch": 0.08169102853125833,
      "grad_norm": 896.7678833007812,
      "learning_rate": 4.084848484848485e-05,
      "loss": 171.1982,
      "step": 10110
    },
    {
      "epoch": 0.08177183073554246,
      "grad_norm": 2560.443115234375,
      "learning_rate": 4.088888888888889e-05,
      "loss": 242.9725,
      "step": 10120
    },
    {
      "epoch": 0.0818526329398266,
      "grad_norm": 914.347900390625,
      "learning_rate": 4.092929292929293e-05,
      "loss": 166.4831,
      "step": 10130
    },
    {
      "epoch": 0.08193343514411074,
      "grad_norm": 1105.142822265625,
      "learning_rate": 4.096969696969697e-05,
      "loss": 182.9737,
      "step": 10140
    },
    {
      "epoch": 0.08201423734839486,
      "grad_norm": 1321.1329345703125,
      "learning_rate": 4.101010101010101e-05,
      "loss": 152.9314,
      "step": 10150
    },
    {
      "epoch": 0.082095039552679,
      "grad_norm": 924.438720703125,
      "learning_rate": 4.105050505050505e-05,
      "loss": 187.9318,
      "step": 10160
    },
    {
      "epoch": 0.08217584175696313,
      "grad_norm": 834.27685546875,
      "learning_rate": 4.109090909090909e-05,
      "loss": 177.3875,
      "step": 10170
    },
    {
      "epoch": 0.08225664396124727,
      "grad_norm": 996.1378173828125,
      "learning_rate": 4.113131313131313e-05,
      "loss": 163.1871,
      "step": 10180
    },
    {
      "epoch": 0.08233744616553139,
      "grad_norm": 1657.6314697265625,
      "learning_rate": 4.117171717171717e-05,
      "loss": 195.5306,
      "step": 10190
    },
    {
      "epoch": 0.08241824836981552,
      "grad_norm": 1040.5526123046875,
      "learning_rate": 4.1212121212121216e-05,
      "loss": 179.7722,
      "step": 10200
    },
    {
      "epoch": 0.08249905057409966,
      "grad_norm": 1405.0408935546875,
      "learning_rate": 4.1252525252525255e-05,
      "loss": 177.9881,
      "step": 10210
    },
    {
      "epoch": 0.0825798527783838,
      "grad_norm": 1484.392333984375,
      "learning_rate": 4.1292929292929294e-05,
      "loss": 170.4384,
      "step": 10220
    },
    {
      "epoch": 0.08266065498266793,
      "grad_norm": 533.9537963867188,
      "learning_rate": 4.133333333333333e-05,
      "loss": 166.7928,
      "step": 10230
    },
    {
      "epoch": 0.08274145718695206,
      "grad_norm": 1133.5531005859375,
      "learning_rate": 4.137373737373738e-05,
      "loss": 185.8696,
      "step": 10240
    },
    {
      "epoch": 0.08282225939123619,
      "grad_norm": 1964.5546875,
      "learning_rate": 4.141414141414142e-05,
      "loss": 247.6705,
      "step": 10250
    },
    {
      "epoch": 0.08290306159552033,
      "grad_norm": 1816.8203125,
      "learning_rate": 4.1454545454545456e-05,
      "loss": 188.6205,
      "step": 10260
    },
    {
      "epoch": 0.08298386379980446,
      "grad_norm": 809.494873046875,
      "learning_rate": 4.1494949494949495e-05,
      "loss": 186.0191,
      "step": 10270
    },
    {
      "epoch": 0.08306466600408859,
      "grad_norm": 1444.1771240234375,
      "learning_rate": 4.153535353535354e-05,
      "loss": 171.1266,
      "step": 10280
    },
    {
      "epoch": 0.08314546820837272,
      "grad_norm": 1594.9212646484375,
      "learning_rate": 4.157575757575758e-05,
      "loss": 152.8558,
      "step": 10290
    },
    {
      "epoch": 0.08322627041265686,
      "grad_norm": 1367.26318359375,
      "learning_rate": 4.161616161616162e-05,
      "loss": 215.3903,
      "step": 10300
    },
    {
      "epoch": 0.083307072616941,
      "grad_norm": 1395.453857421875,
      "learning_rate": 4.165656565656566e-05,
      "loss": 219.8608,
      "step": 10310
    },
    {
      "epoch": 0.08338787482122512,
      "grad_norm": 823.7109375,
      "learning_rate": 4.16969696969697e-05,
      "loss": 198.1472,
      "step": 10320
    },
    {
      "epoch": 0.08346867702550925,
      "grad_norm": 4030.11083984375,
      "learning_rate": 4.173737373737374e-05,
      "loss": 191.6869,
      "step": 10330
    },
    {
      "epoch": 0.08354947922979339,
      "grad_norm": 1047.6395263671875,
      "learning_rate": 4.177777777777778e-05,
      "loss": 198.535,
      "step": 10340
    },
    {
      "epoch": 0.08363028143407752,
      "grad_norm": 1213.222412109375,
      "learning_rate": 4.181818181818182e-05,
      "loss": 209.4711,
      "step": 10350
    },
    {
      "epoch": 0.08371108363836166,
      "grad_norm": 862.3009643554688,
      "learning_rate": 4.185858585858586e-05,
      "loss": 197.0447,
      "step": 10360
    },
    {
      "epoch": 0.08379188584264578,
      "grad_norm": 7555.03271484375,
      "learning_rate": 4.18989898989899e-05,
      "loss": 257.4577,
      "step": 10370
    },
    {
      "epoch": 0.08387268804692992,
      "grad_norm": 1448.51806640625,
      "learning_rate": 4.193939393939394e-05,
      "loss": 145.9192,
      "step": 10380
    },
    {
      "epoch": 0.08395349025121406,
      "grad_norm": 563.4491577148438,
      "learning_rate": 4.197979797979798e-05,
      "loss": 185.379,
      "step": 10390
    },
    {
      "epoch": 0.08403429245549819,
      "grad_norm": 2999.943603515625,
      "learning_rate": 4.202020202020202e-05,
      "loss": 207.6808,
      "step": 10400
    },
    {
      "epoch": 0.08411509465978231,
      "grad_norm": 1272.3822021484375,
      "learning_rate": 4.206060606060606e-05,
      "loss": 203.2208,
      "step": 10410
    },
    {
      "epoch": 0.08419589686406645,
      "grad_norm": 1877.1287841796875,
      "learning_rate": 4.2101010101010105e-05,
      "loss": 208.5024,
      "step": 10420
    },
    {
      "epoch": 0.08427669906835059,
      "grad_norm": 880.4778442382812,
      "learning_rate": 4.2141414141414144e-05,
      "loss": 114.9268,
      "step": 10430
    },
    {
      "epoch": 0.08435750127263472,
      "grad_norm": 2282.030517578125,
      "learning_rate": 4.218181818181818e-05,
      "loss": 185.7249,
      "step": 10440
    },
    {
      "epoch": 0.08443830347691884,
      "grad_norm": 1241.139892578125,
      "learning_rate": 4.222222222222222e-05,
      "loss": 171.4185,
      "step": 10450
    },
    {
      "epoch": 0.08451910568120298,
      "grad_norm": 1106.116943359375,
      "learning_rate": 4.226262626262627e-05,
      "loss": 233.517,
      "step": 10460
    },
    {
      "epoch": 0.08459990788548712,
      "grad_norm": 1352.7723388671875,
      "learning_rate": 4.2303030303030306e-05,
      "loss": 164.0137,
      "step": 10470
    },
    {
      "epoch": 0.08468071008977125,
      "grad_norm": 824.0071411132812,
      "learning_rate": 4.2343434343434345e-05,
      "loss": 149.2571,
      "step": 10480
    },
    {
      "epoch": 0.08476151229405539,
      "grad_norm": 1494.287841796875,
      "learning_rate": 4.2383838383838384e-05,
      "loss": 158.6048,
      "step": 10490
    },
    {
      "epoch": 0.08484231449833951,
      "grad_norm": 928.8883056640625,
      "learning_rate": 4.242424242424243e-05,
      "loss": 160.3929,
      "step": 10500
    },
    {
      "epoch": 0.08492311670262365,
      "grad_norm": 3102.35791015625,
      "learning_rate": 4.246464646464647e-05,
      "loss": 262.7529,
      "step": 10510
    },
    {
      "epoch": 0.08500391890690778,
      "grad_norm": 753.5589599609375,
      "learning_rate": 4.250505050505051e-05,
      "loss": 147.7691,
      "step": 10520
    },
    {
      "epoch": 0.08508472111119192,
      "grad_norm": 1001.4116821289062,
      "learning_rate": 4.254545454545455e-05,
      "loss": 137.3231,
      "step": 10530
    },
    {
      "epoch": 0.08516552331547604,
      "grad_norm": 813.2144775390625,
      "learning_rate": 4.258585858585859e-05,
      "loss": 201.1259,
      "step": 10540
    },
    {
      "epoch": 0.08524632551976018,
      "grad_norm": 1098.597900390625,
      "learning_rate": 4.262626262626263e-05,
      "loss": 165.8937,
      "step": 10550
    },
    {
      "epoch": 0.08532712772404431,
      "grad_norm": 1298.8853759765625,
      "learning_rate": 4.266666666666667e-05,
      "loss": 198.1985,
      "step": 10560
    },
    {
      "epoch": 0.08540792992832845,
      "grad_norm": 1016.0570678710938,
      "learning_rate": 4.270707070707071e-05,
      "loss": 206.1271,
      "step": 10570
    },
    {
      "epoch": 0.08548873213261257,
      "grad_norm": 2057.573974609375,
      "learning_rate": 4.274747474747475e-05,
      "loss": 172.7861,
      "step": 10580
    },
    {
      "epoch": 0.08556953433689671,
      "grad_norm": 1261.774169921875,
      "learning_rate": 4.2787878787878786e-05,
      "loss": 149.1268,
      "step": 10590
    },
    {
      "epoch": 0.08565033654118084,
      "grad_norm": 1344.2037353515625,
      "learning_rate": 4.282828282828283e-05,
      "loss": 195.2029,
      "step": 10600
    },
    {
      "epoch": 0.08573113874546498,
      "grad_norm": 3034.935546875,
      "learning_rate": 4.286868686868687e-05,
      "loss": 222.7399,
      "step": 10610
    },
    {
      "epoch": 0.08581194094974912,
      "grad_norm": 970.0159912109375,
      "learning_rate": 4.290909090909091e-05,
      "loss": 168.5838,
      "step": 10620
    },
    {
      "epoch": 0.08589274315403324,
      "grad_norm": 953.5883178710938,
      "learning_rate": 4.294949494949495e-05,
      "loss": 162.8468,
      "step": 10630
    },
    {
      "epoch": 0.08597354535831737,
      "grad_norm": 2465.924072265625,
      "learning_rate": 4.2989898989898994e-05,
      "loss": 204.2057,
      "step": 10640
    },
    {
      "epoch": 0.08605434756260151,
      "grad_norm": 1194.1285400390625,
      "learning_rate": 4.303030303030303e-05,
      "loss": 197.6671,
      "step": 10650
    },
    {
      "epoch": 0.08613514976688565,
      "grad_norm": 881.1695556640625,
      "learning_rate": 4.307070707070707e-05,
      "loss": 163.8398,
      "step": 10660
    },
    {
      "epoch": 0.08621595197116977,
      "grad_norm": 3846.036376953125,
      "learning_rate": 4.311111111111111e-05,
      "loss": 178.4458,
      "step": 10670
    },
    {
      "epoch": 0.0862967541754539,
      "grad_norm": 1206.0906982421875,
      "learning_rate": 4.3151515151515156e-05,
      "loss": 153.2348,
      "step": 10680
    },
    {
      "epoch": 0.08637755637973804,
      "grad_norm": 898.4664306640625,
      "learning_rate": 4.3191919191919195e-05,
      "loss": 143.3541,
      "step": 10690
    },
    {
      "epoch": 0.08645835858402218,
      "grad_norm": 2398.2255859375,
      "learning_rate": 4.3232323232323234e-05,
      "loss": 212.5595,
      "step": 10700
    },
    {
      "epoch": 0.0865391607883063,
      "grad_norm": 1220.5733642578125,
      "learning_rate": 4.327272727272728e-05,
      "loss": 179.1572,
      "step": 10710
    },
    {
      "epoch": 0.08661996299259044,
      "grad_norm": 1974.5457763671875,
      "learning_rate": 4.331313131313132e-05,
      "loss": 190.0654,
      "step": 10720
    },
    {
      "epoch": 0.08670076519687457,
      "grad_norm": 2125.98583984375,
      "learning_rate": 4.335353535353536e-05,
      "loss": 168.3283,
      "step": 10730
    },
    {
      "epoch": 0.08678156740115871,
      "grad_norm": 1389.6546630859375,
      "learning_rate": 4.3393939393939396e-05,
      "loss": 159.5211,
      "step": 10740
    },
    {
      "epoch": 0.08686236960544283,
      "grad_norm": 807.5780029296875,
      "learning_rate": 4.343434343434344e-05,
      "loss": 192.4806,
      "step": 10750
    },
    {
      "epoch": 0.08694317180972697,
      "grad_norm": 1139.113037109375,
      "learning_rate": 4.347474747474748e-05,
      "loss": 155.3915,
      "step": 10760
    },
    {
      "epoch": 0.0870239740140111,
      "grad_norm": 1003.0131225585938,
      "learning_rate": 4.351515151515152e-05,
      "loss": 150.0795,
      "step": 10770
    },
    {
      "epoch": 0.08710477621829524,
      "grad_norm": 962.0811767578125,
      "learning_rate": 4.355555555555556e-05,
      "loss": 194.0887,
      "step": 10780
    },
    {
      "epoch": 0.08718557842257937,
      "grad_norm": 626.3510131835938,
      "learning_rate": 4.35959595959596e-05,
      "loss": 148.0051,
      "step": 10790
    },
    {
      "epoch": 0.0872663806268635,
      "grad_norm": 2664.0517578125,
      "learning_rate": 4.3636363636363636e-05,
      "loss": 199.106,
      "step": 10800
    },
    {
      "epoch": 0.08734718283114763,
      "grad_norm": 895.5698852539062,
      "learning_rate": 4.3676767676767674e-05,
      "loss": 172.3583,
      "step": 10810
    },
    {
      "epoch": 0.08742798503543177,
      "grad_norm": 1103.621826171875,
      "learning_rate": 4.371717171717172e-05,
      "loss": 203.0746,
      "step": 10820
    },
    {
      "epoch": 0.0875087872397159,
      "grad_norm": 1323.6517333984375,
      "learning_rate": 4.375757575757576e-05,
      "loss": 158.2676,
      "step": 10830
    },
    {
      "epoch": 0.08758958944400003,
      "grad_norm": 1170.48779296875,
      "learning_rate": 4.37979797979798e-05,
      "loss": 151.4179,
      "step": 10840
    },
    {
      "epoch": 0.08767039164828416,
      "grad_norm": 1171.2379150390625,
      "learning_rate": 4.383838383838384e-05,
      "loss": 181.4644,
      "step": 10850
    },
    {
      "epoch": 0.0877511938525683,
      "grad_norm": 1495.01025390625,
      "learning_rate": 4.387878787878788e-05,
      "loss": 167.6518,
      "step": 10860
    },
    {
      "epoch": 0.08783199605685243,
      "grad_norm": 1283.5498046875,
      "learning_rate": 4.391919191919192e-05,
      "loss": 138.4191,
      "step": 10870
    },
    {
      "epoch": 0.08791279826113656,
      "grad_norm": 1028.198974609375,
      "learning_rate": 4.395959595959596e-05,
      "loss": 167.4624,
      "step": 10880
    },
    {
      "epoch": 0.0879936004654207,
      "grad_norm": 958.3167114257812,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 187.0603,
      "step": 10890
    },
    {
      "epoch": 0.08807440266970483,
      "grad_norm": 1248.95556640625,
      "learning_rate": 4.4040404040404044e-05,
      "loss": 194.5838,
      "step": 10900
    },
    {
      "epoch": 0.08815520487398897,
      "grad_norm": 1088.8775634765625,
      "learning_rate": 4.408080808080808e-05,
      "loss": 137.5682,
      "step": 10910
    },
    {
      "epoch": 0.0882360070782731,
      "grad_norm": 1130.275146484375,
      "learning_rate": 4.412121212121212e-05,
      "loss": 180.1215,
      "step": 10920
    },
    {
      "epoch": 0.08831680928255722,
      "grad_norm": 1201.9453125,
      "learning_rate": 4.416161616161617e-05,
      "loss": 157.9624,
      "step": 10930
    },
    {
      "epoch": 0.08839761148684136,
      "grad_norm": 1291.0989990234375,
      "learning_rate": 4.420202020202021e-05,
      "loss": 159.1208,
      "step": 10940
    },
    {
      "epoch": 0.0884784136911255,
      "grad_norm": 946.185546875,
      "learning_rate": 4.4242424242424246e-05,
      "loss": 200.6266,
      "step": 10950
    },
    {
      "epoch": 0.08855921589540963,
      "grad_norm": 2330.45361328125,
      "learning_rate": 4.4282828282828284e-05,
      "loss": 163.2997,
      "step": 10960
    },
    {
      "epoch": 0.08864001809969375,
      "grad_norm": 959.3818359375,
      "learning_rate": 4.432323232323233e-05,
      "loss": 200.18,
      "step": 10970
    },
    {
      "epoch": 0.08872082030397789,
      "grad_norm": 1215.0078125,
      "learning_rate": 4.436363636363637e-05,
      "loss": 194.7184,
      "step": 10980
    },
    {
      "epoch": 0.08880162250826203,
      "grad_norm": 731.89501953125,
      "learning_rate": 4.44040404040404e-05,
      "loss": 183.9504,
      "step": 10990
    },
    {
      "epoch": 0.08888242471254616,
      "grad_norm": 751.2623291015625,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 164.3441,
      "step": 11000
    },
    {
      "epoch": 0.08896322691683028,
      "grad_norm": 1424.0635986328125,
      "learning_rate": 4.4484848484848485e-05,
      "loss": 202.2657,
      "step": 11010
    },
    {
      "epoch": 0.08904402912111442,
      "grad_norm": 1039.608642578125,
      "learning_rate": 4.4525252525252524e-05,
      "loss": 177.8795,
      "step": 11020
    },
    {
      "epoch": 0.08912483132539856,
      "grad_norm": 1338.993408203125,
      "learning_rate": 4.456565656565656e-05,
      "loss": 167.3348,
      "step": 11030
    },
    {
      "epoch": 0.08920563352968269,
      "grad_norm": 2591.8984375,
      "learning_rate": 4.460606060606061e-05,
      "loss": 165.2155,
      "step": 11040
    },
    {
      "epoch": 0.08928643573396683,
      "grad_norm": 931.5535888671875,
      "learning_rate": 4.464646464646465e-05,
      "loss": 198.5493,
      "step": 11050
    },
    {
      "epoch": 0.08936723793825095,
      "grad_norm": 1028.25927734375,
      "learning_rate": 4.4686868686868686e-05,
      "loss": 169.1925,
      "step": 11060
    },
    {
      "epoch": 0.08944804014253509,
      "grad_norm": 1156.0810546875,
      "learning_rate": 4.472727272727273e-05,
      "loss": 172.742,
      "step": 11070
    },
    {
      "epoch": 0.08952884234681922,
      "grad_norm": 1409.868408203125,
      "learning_rate": 4.476767676767677e-05,
      "loss": 152.6085,
      "step": 11080
    },
    {
      "epoch": 0.08960964455110336,
      "grad_norm": 1091.4266357421875,
      "learning_rate": 4.480808080808081e-05,
      "loss": 149.4833,
      "step": 11090
    },
    {
      "epoch": 0.08969044675538748,
      "grad_norm": 1274.9849853515625,
      "learning_rate": 4.484848484848485e-05,
      "loss": 239.724,
      "step": 11100
    },
    {
      "epoch": 0.08977124895967162,
      "grad_norm": 2974.341552734375,
      "learning_rate": 4.4888888888888894e-05,
      "loss": 184.0551,
      "step": 11110
    },
    {
      "epoch": 0.08985205116395575,
      "grad_norm": 893.275390625,
      "learning_rate": 4.492929292929293e-05,
      "loss": 163.8494,
      "step": 11120
    },
    {
      "epoch": 0.08993285336823989,
      "grad_norm": 1236.2047119140625,
      "learning_rate": 4.496969696969697e-05,
      "loss": 223.2573,
      "step": 11130
    },
    {
      "epoch": 0.09001365557252401,
      "grad_norm": 560.5221557617188,
      "learning_rate": 4.501010101010101e-05,
      "loss": 143.0442,
      "step": 11140
    },
    {
      "epoch": 0.09009445777680815,
      "grad_norm": 1671.4537353515625,
      "learning_rate": 4.5050505050505056e-05,
      "loss": 230.4551,
      "step": 11150
    },
    {
      "epoch": 0.09017525998109228,
      "grad_norm": 2295.0419921875,
      "learning_rate": 4.5090909090909095e-05,
      "loss": 191.5067,
      "step": 11160
    },
    {
      "epoch": 0.09025606218537642,
      "grad_norm": 1291.7230224609375,
      "learning_rate": 4.5131313131313134e-05,
      "loss": 165.677,
      "step": 11170
    },
    {
      "epoch": 0.09033686438966056,
      "grad_norm": 817.165771484375,
      "learning_rate": 4.517171717171717e-05,
      "loss": 145.0678,
      "step": 11180
    },
    {
      "epoch": 0.09041766659394468,
      "grad_norm": 938.4746704101562,
      "learning_rate": 4.521212121212122e-05,
      "loss": 168.7315,
      "step": 11190
    },
    {
      "epoch": 0.09049846879822881,
      "grad_norm": 820.7261352539062,
      "learning_rate": 4.525252525252526e-05,
      "loss": 174.4509,
      "step": 11200
    },
    {
      "epoch": 0.09057927100251295,
      "grad_norm": 899.671875,
      "learning_rate": 4.529292929292929e-05,
      "loss": 153.883,
      "step": 11210
    },
    {
      "epoch": 0.09066007320679709,
      "grad_norm": 2744.694091796875,
      "learning_rate": 4.5333333333333335e-05,
      "loss": 152.8033,
      "step": 11220
    },
    {
      "epoch": 0.09074087541108121,
      "grad_norm": 1821.427734375,
      "learning_rate": 4.5373737373737374e-05,
      "loss": 203.8027,
      "step": 11230
    },
    {
      "epoch": 0.09082167761536535,
      "grad_norm": 937.9207763671875,
      "learning_rate": 4.541414141414141e-05,
      "loss": 183.0782,
      "step": 11240
    },
    {
      "epoch": 0.09090247981964948,
      "grad_norm": 1872.185546875,
      "learning_rate": 4.545454545454546e-05,
      "loss": 153.9125,
      "step": 11250
    },
    {
      "epoch": 0.09098328202393362,
      "grad_norm": 772.832275390625,
      "learning_rate": 4.54949494949495e-05,
      "loss": 143.9607,
      "step": 11260
    },
    {
      "epoch": 0.09106408422821774,
      "grad_norm": 2842.652587890625,
      "learning_rate": 4.5535353535353536e-05,
      "loss": 189.6786,
      "step": 11270
    },
    {
      "epoch": 0.09114488643250188,
      "grad_norm": 1738.2589111328125,
      "learning_rate": 4.5575757575757575e-05,
      "loss": 163.1939,
      "step": 11280
    },
    {
      "epoch": 0.09122568863678601,
      "grad_norm": 1521.9814453125,
      "learning_rate": 4.561616161616162e-05,
      "loss": 208.5482,
      "step": 11290
    },
    {
      "epoch": 0.09130649084107015,
      "grad_norm": 1132.692138671875,
      "learning_rate": 4.565656565656566e-05,
      "loss": 161.8789,
      "step": 11300
    },
    {
      "epoch": 0.09138729304535427,
      "grad_norm": 1395.38671875,
      "learning_rate": 4.56969696969697e-05,
      "loss": 164.5156,
      "step": 11310
    },
    {
      "epoch": 0.0914680952496384,
      "grad_norm": 661.7669067382812,
      "learning_rate": 4.573737373737374e-05,
      "loss": 163.1058,
      "step": 11320
    },
    {
      "epoch": 0.09154889745392254,
      "grad_norm": 1353.42578125,
      "learning_rate": 4.577777777777778e-05,
      "loss": 178.0265,
      "step": 11330
    },
    {
      "epoch": 0.09162969965820668,
      "grad_norm": 758.9215087890625,
      "learning_rate": 4.581818181818182e-05,
      "loss": 186.8539,
      "step": 11340
    },
    {
      "epoch": 0.09171050186249081,
      "grad_norm": 926.5440673828125,
      "learning_rate": 4.585858585858586e-05,
      "loss": 174.9289,
      "step": 11350
    },
    {
      "epoch": 0.09179130406677494,
      "grad_norm": 1032.7493896484375,
      "learning_rate": 4.58989898989899e-05,
      "loss": 157.8875,
      "step": 11360
    },
    {
      "epoch": 0.09187210627105907,
      "grad_norm": 1104.685302734375,
      "learning_rate": 4.5939393939393945e-05,
      "loss": 180.8059,
      "step": 11370
    },
    {
      "epoch": 0.09195290847534321,
      "grad_norm": 807.3258056640625,
      "learning_rate": 4.5979797979797984e-05,
      "loss": 133.7691,
      "step": 11380
    },
    {
      "epoch": 0.09203371067962735,
      "grad_norm": 1296.3505859375,
      "learning_rate": 4.602020202020202e-05,
      "loss": 168.8188,
      "step": 11390
    },
    {
      "epoch": 0.09211451288391147,
      "grad_norm": 717.891357421875,
      "learning_rate": 4.606060606060607e-05,
      "loss": 143.9607,
      "step": 11400
    },
    {
      "epoch": 0.0921953150881956,
      "grad_norm": 841.2793579101562,
      "learning_rate": 4.610101010101011e-05,
      "loss": 185.6871,
      "step": 11410
    },
    {
      "epoch": 0.09227611729247974,
      "grad_norm": 1319.2064208984375,
      "learning_rate": 4.614141414141414e-05,
      "loss": 163.8516,
      "step": 11420
    },
    {
      "epoch": 0.09235691949676388,
      "grad_norm": 1566.4168701171875,
      "learning_rate": 4.618181818181818e-05,
      "loss": 165.3434,
      "step": 11430
    },
    {
      "epoch": 0.092437721701048,
      "grad_norm": 919.2448120117188,
      "learning_rate": 4.6222222222222224e-05,
      "loss": 133.6576,
      "step": 11440
    },
    {
      "epoch": 0.09251852390533213,
      "grad_norm": 976.5399780273438,
      "learning_rate": 4.626262626262626e-05,
      "loss": 172.4401,
      "step": 11450
    },
    {
      "epoch": 0.09259932610961627,
      "grad_norm": 1343.291015625,
      "learning_rate": 4.63030303030303e-05,
      "loss": 203.6782,
      "step": 11460
    },
    {
      "epoch": 0.0926801283139004,
      "grad_norm": 839.6242065429688,
      "learning_rate": 4.634343434343435e-05,
      "loss": 175.6062,
      "step": 11470
    },
    {
      "epoch": 0.09276093051818454,
      "grad_norm": 909.182861328125,
      "learning_rate": 4.6383838383838386e-05,
      "loss": 159.7895,
      "step": 11480
    },
    {
      "epoch": 0.09284173272246866,
      "grad_norm": 2260.60107421875,
      "learning_rate": 4.6424242424242425e-05,
      "loss": 168.3035,
      "step": 11490
    },
    {
      "epoch": 0.0929225349267528,
      "grad_norm": 1016.5995483398438,
      "learning_rate": 4.6464646464646464e-05,
      "loss": 149.6134,
      "step": 11500
    },
    {
      "epoch": 0.09300333713103694,
      "grad_norm": 1180.2608642578125,
      "learning_rate": 4.650505050505051e-05,
      "loss": 123.0571,
      "step": 11510
    },
    {
      "epoch": 0.09308413933532107,
      "grad_norm": 1054.5244140625,
      "learning_rate": 4.654545454545455e-05,
      "loss": 180.0858,
      "step": 11520
    },
    {
      "epoch": 0.0931649415396052,
      "grad_norm": 1064.1981201171875,
      "learning_rate": 4.658585858585859e-05,
      "loss": 305.0233,
      "step": 11530
    },
    {
      "epoch": 0.09324574374388933,
      "grad_norm": 1455.9857177734375,
      "learning_rate": 4.6626262626262626e-05,
      "loss": 174.1713,
      "step": 11540
    },
    {
      "epoch": 0.09332654594817347,
      "grad_norm": 801.254150390625,
      "learning_rate": 4.666666666666667e-05,
      "loss": 175.7632,
      "step": 11550
    },
    {
      "epoch": 0.0934073481524576,
      "grad_norm": 811.2909545898438,
      "learning_rate": 4.670707070707071e-05,
      "loss": 152.5526,
      "step": 11560
    },
    {
      "epoch": 0.09348815035674173,
      "grad_norm": 860.8350830078125,
      "learning_rate": 4.674747474747475e-05,
      "loss": 159.9338,
      "step": 11570
    },
    {
      "epoch": 0.09356895256102586,
      "grad_norm": 1820.26318359375,
      "learning_rate": 4.6787878787878795e-05,
      "loss": 205.222,
      "step": 11580
    },
    {
      "epoch": 0.09364975476531,
      "grad_norm": 1027.9521484375,
      "learning_rate": 4.6828282828282834e-05,
      "loss": 167.9507,
      "step": 11590
    },
    {
      "epoch": 0.09373055696959413,
      "grad_norm": 3551.64599609375,
      "learning_rate": 4.686868686868687e-05,
      "loss": 194.2529,
      "step": 11600
    },
    {
      "epoch": 0.09381135917387827,
      "grad_norm": 957.3357543945312,
      "learning_rate": 4.690909090909091e-05,
      "loss": 185.5651,
      "step": 11610
    },
    {
      "epoch": 0.09389216137816239,
      "grad_norm": 587.98828125,
      "learning_rate": 4.694949494949496e-05,
      "loss": 208.2654,
      "step": 11620
    },
    {
      "epoch": 0.09397296358244653,
      "grad_norm": 824.1953735351562,
      "learning_rate": 4.698989898989899e-05,
      "loss": 118.9412,
      "step": 11630
    },
    {
      "epoch": 0.09405376578673066,
      "grad_norm": 997.8128051757812,
      "learning_rate": 4.703030303030303e-05,
      "loss": 144.5133,
      "step": 11640
    },
    {
      "epoch": 0.0941345679910148,
      "grad_norm": 825.6588745117188,
      "learning_rate": 4.7070707070707074e-05,
      "loss": 130.3822,
      "step": 11650
    },
    {
      "epoch": 0.09421537019529892,
      "grad_norm": 1590.2271728515625,
      "learning_rate": 4.711111111111111e-05,
      "loss": 147.783,
      "step": 11660
    },
    {
      "epoch": 0.09429617239958306,
      "grad_norm": 831.8695068359375,
      "learning_rate": 4.715151515151515e-05,
      "loss": 150.0897,
      "step": 11670
    },
    {
      "epoch": 0.0943769746038672,
      "grad_norm": 879.4678955078125,
      "learning_rate": 4.719191919191919e-05,
      "loss": 134.1156,
      "step": 11680
    },
    {
      "epoch": 0.09445777680815133,
      "grad_norm": 1955.2484130859375,
      "learning_rate": 4.7232323232323236e-05,
      "loss": 155.8984,
      "step": 11690
    },
    {
      "epoch": 0.09453857901243545,
      "grad_norm": 1074.5509033203125,
      "learning_rate": 4.7272727272727275e-05,
      "loss": 186.7514,
      "step": 11700
    },
    {
      "epoch": 0.09461938121671959,
      "grad_norm": 992.748046875,
      "learning_rate": 4.7313131313131314e-05,
      "loss": 155.1787,
      "step": 11710
    },
    {
      "epoch": 0.09470018342100373,
      "grad_norm": 990.5753784179688,
      "learning_rate": 4.735353535353535e-05,
      "loss": 140.0309,
      "step": 11720
    },
    {
      "epoch": 0.09478098562528786,
      "grad_norm": 792.1434936523438,
      "learning_rate": 4.73939393939394e-05,
      "loss": 138.4607,
      "step": 11730
    },
    {
      "epoch": 0.094861787829572,
      "grad_norm": 1121.582763671875,
      "learning_rate": 4.743434343434344e-05,
      "loss": 121.2844,
      "step": 11740
    },
    {
      "epoch": 0.09494259003385612,
      "grad_norm": 1351.6878662109375,
      "learning_rate": 4.7474747474747476e-05,
      "loss": 198.7702,
      "step": 11750
    },
    {
      "epoch": 0.09502339223814026,
      "grad_norm": 2031.31494140625,
      "learning_rate": 4.751515151515152e-05,
      "loss": 164.1413,
      "step": 11760
    },
    {
      "epoch": 0.09510419444242439,
      "grad_norm": 761.006103515625,
      "learning_rate": 4.755555555555556e-05,
      "loss": 143.7779,
      "step": 11770
    },
    {
      "epoch": 0.09518499664670853,
      "grad_norm": 1228.6676025390625,
      "learning_rate": 4.75959595959596e-05,
      "loss": 177.4899,
      "step": 11780
    },
    {
      "epoch": 0.09526579885099265,
      "grad_norm": 817.8377075195312,
      "learning_rate": 4.763636363636364e-05,
      "loss": 158.3827,
      "step": 11790
    },
    {
      "epoch": 0.09534660105527679,
      "grad_norm": 1481.2501220703125,
      "learning_rate": 4.7676767676767684e-05,
      "loss": 144.1627,
      "step": 11800
    },
    {
      "epoch": 0.09542740325956092,
      "grad_norm": 899.5111083984375,
      "learning_rate": 4.771717171717172e-05,
      "loss": 188.1043,
      "step": 11810
    },
    {
      "epoch": 0.09550820546384506,
      "grad_norm": 766.15869140625,
      "learning_rate": 4.775757575757576e-05,
      "loss": 190.8986,
      "step": 11820
    },
    {
      "epoch": 0.09558900766812918,
      "grad_norm": 818.2703857421875,
      "learning_rate": 4.77979797979798e-05,
      "loss": 226.6272,
      "step": 11830
    },
    {
      "epoch": 0.09566980987241332,
      "grad_norm": 1306.8607177734375,
      "learning_rate": 4.7838383838383846e-05,
      "loss": 179.7575,
      "step": 11840
    },
    {
      "epoch": 0.09575061207669745,
      "grad_norm": 1371.1048583984375,
      "learning_rate": 4.787878787878788e-05,
      "loss": 184.5432,
      "step": 11850
    },
    {
      "epoch": 0.09583141428098159,
      "grad_norm": 1219.8428955078125,
      "learning_rate": 4.791919191919192e-05,
      "loss": 256.0484,
      "step": 11860
    },
    {
      "epoch": 0.09591221648526572,
      "grad_norm": 1769.408935546875,
      "learning_rate": 4.795959595959596e-05,
      "loss": 216.8512,
      "step": 11870
    },
    {
      "epoch": 0.09599301868954985,
      "grad_norm": 1021.0985107421875,
      "learning_rate": 4.8e-05,
      "loss": 150.8418,
      "step": 11880
    },
    {
      "epoch": 0.09607382089383398,
      "grad_norm": 789.3172607421875,
      "learning_rate": 4.804040404040404e-05,
      "loss": 168.713,
      "step": 11890
    },
    {
      "epoch": 0.09615462309811812,
      "grad_norm": 1145.4168701171875,
      "learning_rate": 4.808080808080808e-05,
      "loss": 248.8971,
      "step": 11900
    },
    {
      "epoch": 0.09623542530240226,
      "grad_norm": 1330.3175048828125,
      "learning_rate": 4.8121212121212125e-05,
      "loss": 143.9111,
      "step": 11910
    },
    {
      "epoch": 0.09631622750668638,
      "grad_norm": 1145.8402099609375,
      "learning_rate": 4.8161616161616163e-05,
      "loss": 196.2298,
      "step": 11920
    },
    {
      "epoch": 0.09639702971097051,
      "grad_norm": 3719.409423828125,
      "learning_rate": 4.82020202020202e-05,
      "loss": 141.981,
      "step": 11930
    },
    {
      "epoch": 0.09647783191525465,
      "grad_norm": 783.9396362304688,
      "learning_rate": 4.824242424242425e-05,
      "loss": 139.5111,
      "step": 11940
    },
    {
      "epoch": 0.09655863411953879,
      "grad_norm": 817.3587036132812,
      "learning_rate": 4.828282828282829e-05,
      "loss": 193.1281,
      "step": 11950
    },
    {
      "epoch": 0.09663943632382291,
      "grad_norm": 2150.12939453125,
      "learning_rate": 4.8323232323232326e-05,
      "loss": 173.1304,
      "step": 11960
    },
    {
      "epoch": 0.09672023852810704,
      "grad_norm": 1299.8162841796875,
      "learning_rate": 4.8363636363636364e-05,
      "loss": 181.7705,
      "step": 11970
    },
    {
      "epoch": 0.09680104073239118,
      "grad_norm": 936.7069091796875,
      "learning_rate": 4.840404040404041e-05,
      "loss": 173.4242,
      "step": 11980
    },
    {
      "epoch": 0.09688184293667532,
      "grad_norm": 613.501708984375,
      "learning_rate": 4.844444444444445e-05,
      "loss": 153.117,
      "step": 11990
    },
    {
      "epoch": 0.09696264514095944,
      "grad_norm": 1406.6314697265625,
      "learning_rate": 4.848484848484849e-05,
      "loss": 185.4404,
      "step": 12000
    },
    {
      "epoch": 0.09704344734524357,
      "grad_norm": 1215.42138671875,
      "learning_rate": 4.852525252525253e-05,
      "loss": 209.0183,
      "step": 12010
    },
    {
      "epoch": 0.09712424954952771,
      "grad_norm": 1375.722412109375,
      "learning_rate": 4.856565656565657e-05,
      "loss": 201.9598,
      "step": 12020
    },
    {
      "epoch": 0.09720505175381185,
      "grad_norm": 987.8963623046875,
      "learning_rate": 4.860606060606061e-05,
      "loss": 151.2616,
      "step": 12030
    },
    {
      "epoch": 0.09728585395809598,
      "grad_norm": 1280.0211181640625,
      "learning_rate": 4.864646464646465e-05,
      "loss": 137.8948,
      "step": 12040
    },
    {
      "epoch": 0.0973666561623801,
      "grad_norm": 1247.3785400390625,
      "learning_rate": 4.868686868686869e-05,
      "loss": 154.3858,
      "step": 12050
    },
    {
      "epoch": 0.09744745836666424,
      "grad_norm": 1069.8919677734375,
      "learning_rate": 4.872727272727273e-05,
      "loss": 154.6211,
      "step": 12060
    },
    {
      "epoch": 0.09752826057094838,
      "grad_norm": 1880.9874267578125,
      "learning_rate": 4.8767676767676767e-05,
      "loss": 149.5197,
      "step": 12070
    },
    {
      "epoch": 0.09760906277523251,
      "grad_norm": 1230.8167724609375,
      "learning_rate": 4.8808080808080805e-05,
      "loss": 149.4375,
      "step": 12080
    },
    {
      "epoch": 0.09768986497951664,
      "grad_norm": 1191.7757568359375,
      "learning_rate": 4.884848484848485e-05,
      "loss": 170.0373,
      "step": 12090
    },
    {
      "epoch": 0.09777066718380077,
      "grad_norm": 1369.3248291015625,
      "learning_rate": 4.888888888888889e-05,
      "loss": 161.5461,
      "step": 12100
    },
    {
      "epoch": 0.09785146938808491,
      "grad_norm": 864.84423828125,
      "learning_rate": 4.892929292929293e-05,
      "loss": 185.1635,
      "step": 12110
    },
    {
      "epoch": 0.09793227159236904,
      "grad_norm": 1059.955810546875,
      "learning_rate": 4.896969696969697e-05,
      "loss": 190.465,
      "step": 12120
    },
    {
      "epoch": 0.09801307379665317,
      "grad_norm": 1468.4659423828125,
      "learning_rate": 4.901010101010101e-05,
      "loss": 209.5102,
      "step": 12130
    },
    {
      "epoch": 0.0980938760009373,
      "grad_norm": 1922.64404296875,
      "learning_rate": 4.905050505050505e-05,
      "loss": 136.4074,
      "step": 12140
    },
    {
      "epoch": 0.09817467820522144,
      "grad_norm": 731.4076538085938,
      "learning_rate": 4.909090909090909e-05,
      "loss": 132.6805,
      "step": 12150
    },
    {
      "epoch": 0.09825548040950557,
      "grad_norm": 1013.0802612304688,
      "learning_rate": 4.9131313131313137e-05,
      "loss": 169.2464,
      "step": 12160
    },
    {
      "epoch": 0.09833628261378971,
      "grad_norm": 1180.7650146484375,
      "learning_rate": 4.9171717171717175e-05,
      "loss": 134.0749,
      "step": 12170
    },
    {
      "epoch": 0.09841708481807383,
      "grad_norm": 1976.8056640625,
      "learning_rate": 4.9212121212121214e-05,
      "loss": 173.6256,
      "step": 12180
    },
    {
      "epoch": 0.09849788702235797,
      "grad_norm": 1068.174560546875,
      "learning_rate": 4.925252525252525e-05,
      "loss": 172.0746,
      "step": 12190
    },
    {
      "epoch": 0.0985786892266421,
      "grad_norm": 707.3936157226562,
      "learning_rate": 4.92929292929293e-05,
      "loss": 186.8521,
      "step": 12200
    },
    {
      "epoch": 0.09865949143092624,
      "grad_norm": 1194.782470703125,
      "learning_rate": 4.933333333333334e-05,
      "loss": 218.8846,
      "step": 12210
    },
    {
      "epoch": 0.09874029363521036,
      "grad_norm": 1251.564453125,
      "learning_rate": 4.9373737373737376e-05,
      "loss": 163.8027,
      "step": 12220
    },
    {
      "epoch": 0.0988210958394945,
      "grad_norm": 1061.1964111328125,
      "learning_rate": 4.9414141414141415e-05,
      "loss": 189.2458,
      "step": 12230
    },
    {
      "epoch": 0.09890189804377864,
      "grad_norm": 946.7278442382812,
      "learning_rate": 4.945454545454546e-05,
      "loss": 147.6934,
      "step": 12240
    },
    {
      "epoch": 0.09898270024806277,
      "grad_norm": 1294.2952880859375,
      "learning_rate": 4.94949494949495e-05,
      "loss": 224.1921,
      "step": 12250
    },
    {
      "epoch": 0.0990635024523469,
      "grad_norm": 1992.6143798828125,
      "learning_rate": 4.953535353535354e-05,
      "loss": 168.6858,
      "step": 12260
    },
    {
      "epoch": 0.09914430465663103,
      "grad_norm": 931.4608764648438,
      "learning_rate": 4.957575757575758e-05,
      "loss": 164.2498,
      "step": 12270
    },
    {
      "epoch": 0.09922510686091517,
      "grad_norm": 928.4422607421875,
      "learning_rate": 4.9616161616161616e-05,
      "loss": 157.498,
      "step": 12280
    },
    {
      "epoch": 0.0993059090651993,
      "grad_norm": 834.3302612304688,
      "learning_rate": 4.9656565656565655e-05,
      "loss": 195.0842,
      "step": 12290
    },
    {
      "epoch": 0.09938671126948344,
      "grad_norm": 1873.80517578125,
      "learning_rate": 4.9696969696969694e-05,
      "loss": 224.0035,
      "step": 12300
    },
    {
      "epoch": 0.09946751347376756,
      "grad_norm": 1335.8856201171875,
      "learning_rate": 4.973737373737374e-05,
      "loss": 128.0538,
      "step": 12310
    },
    {
      "epoch": 0.0995483156780517,
      "grad_norm": 887.349853515625,
      "learning_rate": 4.977777777777778e-05,
      "loss": 140.0356,
      "step": 12320
    },
    {
      "epoch": 0.09962911788233583,
      "grad_norm": 1496.796630859375,
      "learning_rate": 4.981818181818182e-05,
      "loss": 143.5405,
      "step": 12330
    },
    {
      "epoch": 0.09970992008661997,
      "grad_norm": 1359.33544921875,
      "learning_rate": 4.985858585858586e-05,
      "loss": 163.7442,
      "step": 12340
    },
    {
      "epoch": 0.09979072229090409,
      "grad_norm": 762.21923828125,
      "learning_rate": 4.98989898989899e-05,
      "loss": 159.0385,
      "step": 12350
    },
    {
      "epoch": 0.09987152449518823,
      "grad_norm": 1305.80615234375,
      "learning_rate": 4.993939393939394e-05,
      "loss": 191.1846,
      "step": 12360
    },
    {
      "epoch": 0.09995232669947236,
      "grad_norm": 918.177978515625,
      "learning_rate": 4.997979797979798e-05,
      "loss": 162.5437,
      "step": 12370
    },
    {
      "epoch": 0.1000331289037565,
      "grad_norm": 620.300048828125,
      "learning_rate": 4.9999999751358095e-05,
      "loss": 183.1986,
      "step": 12380
    },
    {
      "epoch": 0.10011393110804062,
      "grad_norm": 1200.64794921875,
      "learning_rate": 4.999999776222285e-05,
      "loss": 125.1974,
      "step": 12390
    },
    {
      "epoch": 0.10019473331232476,
      "grad_norm": 1327.4674072265625,
      "learning_rate": 4.9999993783952516e-05,
      "loss": 203.2498,
      "step": 12400
    },
    {
      "epoch": 0.1002755355166089,
      "grad_norm": 1414.9403076171875,
      "learning_rate": 4.999998781654741e-05,
      "loss": 183.1395,
      "step": 12410
    },
    {
      "epoch": 0.10035633772089303,
      "grad_norm": 1874.2359619140625,
      "learning_rate": 4.9999979860008006e-05,
      "loss": 238.1009,
      "step": 12420
    },
    {
      "epoch": 0.10043713992517717,
      "grad_norm": 1188.5574951171875,
      "learning_rate": 4.9999969914334944e-05,
      "loss": 181.5796,
      "step": 12430
    },
    {
      "epoch": 0.10051794212946129,
      "grad_norm": 1656.1356201171875,
      "learning_rate": 4.999995797952901e-05,
      "loss": 192.1247,
      "step": 12440
    },
    {
      "epoch": 0.10059874433374542,
      "grad_norm": 636.4390869140625,
      "learning_rate": 4.9999944055591154e-05,
      "loss": 144.699,
      "step": 12450
    },
    {
      "epoch": 0.10067954653802956,
      "grad_norm": 958.838134765625,
      "learning_rate": 4.999992814252249e-05,
      "loss": 157.3529,
      "step": 12460
    },
    {
      "epoch": 0.1007603487423137,
      "grad_norm": 1260.2628173828125,
      "learning_rate": 4.999991024032426e-05,
      "loss": 146.3447,
      "step": 12470
    },
    {
      "epoch": 0.10084115094659782,
      "grad_norm": 1077.9141845703125,
      "learning_rate": 4.9999890348997925e-05,
      "loss": 143.8297,
      "step": 12480
    },
    {
      "epoch": 0.10092195315088195,
      "grad_norm": 1348.9508056640625,
      "learning_rate": 4.999986846854504e-05,
      "loss": 173.3844,
      "step": 12490
    },
    {
      "epoch": 0.10100275535516609,
      "grad_norm": 736.1795654296875,
      "learning_rate": 4.999984459896735e-05,
      "loss": 173.8864,
      "step": 12500
    },
    {
      "epoch": 0.10108355755945023,
      "grad_norm": 2159.8544921875,
      "learning_rate": 4.9999818740266766e-05,
      "loss": 262.527,
      "step": 12510
    },
    {
      "epoch": 0.10116435976373435,
      "grad_norm": 1445.3455810546875,
      "learning_rate": 4.999979089244534e-05,
      "loss": 127.1179,
      "step": 12520
    },
    {
      "epoch": 0.10124516196801848,
      "grad_norm": 1050.8997802734375,
      "learning_rate": 4.999976105550528e-05,
      "loss": 171.1639,
      "step": 12530
    },
    {
      "epoch": 0.10132596417230262,
      "grad_norm": 593.4418334960938,
      "learning_rate": 4.9999729229448975e-05,
      "loss": 148.7333,
      "step": 12540
    },
    {
      "epoch": 0.10140676637658676,
      "grad_norm": 1813.0040283203125,
      "learning_rate": 4.9999695414278944e-05,
      "loss": 169.583,
      "step": 12550
    },
    {
      "epoch": 0.10148756858087088,
      "grad_norm": 1632.8077392578125,
      "learning_rate": 4.9999659609997875e-05,
      "loss": 178.7264,
      "step": 12560
    },
    {
      "epoch": 0.10156837078515502,
      "grad_norm": 1241.4422607421875,
      "learning_rate": 4.9999621816608634e-05,
      "loss": 195.7228,
      "step": 12570
    },
    {
      "epoch": 0.10164917298943915,
      "grad_norm": 963.0867309570312,
      "learning_rate": 4.999958203411421e-05,
      "loss": 196.8765,
      "step": 12580
    },
    {
      "epoch": 0.10172997519372329,
      "grad_norm": 879.3704833984375,
      "learning_rate": 4.999954026251778e-05,
      "loss": 172.8657,
      "step": 12590
    },
    {
      "epoch": 0.10181077739800742,
      "grad_norm": 1103.077880859375,
      "learning_rate": 4.999949650182266e-05,
      "loss": 168.6242,
      "step": 12600
    },
    {
      "epoch": 0.10189157960229155,
      "grad_norm": 679.7638549804688,
      "learning_rate": 4.9999450752032345e-05,
      "loss": 130.0992,
      "step": 12610
    },
    {
      "epoch": 0.10197238180657568,
      "grad_norm": 3675.49951171875,
      "learning_rate": 4.999940301315046e-05,
      "loss": 209.7475,
      "step": 12620
    },
    {
      "epoch": 0.10205318401085982,
      "grad_norm": 862.3643188476562,
      "learning_rate": 4.999935328518081e-05,
      "loss": 160.7178,
      "step": 12630
    },
    {
      "epoch": 0.10213398621514395,
      "grad_norm": 894.45068359375,
      "learning_rate": 4.999930156812734e-05,
      "loss": 165.6423,
      "step": 12640
    },
    {
      "epoch": 0.10221478841942808,
      "grad_norm": 856.2412719726562,
      "learning_rate": 4.9999247861994194e-05,
      "loss": 140.6881,
      "step": 12650
    },
    {
      "epoch": 0.10229559062371221,
      "grad_norm": 1286.828125,
      "learning_rate": 4.999919216678561e-05,
      "loss": 177.9921,
      "step": 12660
    },
    {
      "epoch": 0.10237639282799635,
      "grad_norm": 1610.512939453125,
      "learning_rate": 4.999913448250605e-05,
      "loss": 197.9213,
      "step": 12670
    },
    {
      "epoch": 0.10245719503228048,
      "grad_norm": 690.97998046875,
      "learning_rate": 4.999907480916009e-05,
      "loss": 148.6186,
      "step": 12680
    },
    {
      "epoch": 0.10253799723656461,
      "grad_norm": 1106.3350830078125,
      "learning_rate": 4.999901314675246e-05,
      "loss": 209.407,
      "step": 12690
    },
    {
      "epoch": 0.10261879944084874,
      "grad_norm": 1395.5164794921875,
      "learning_rate": 4.99989494952881e-05,
      "loss": 182.4703,
      "step": 12700
    },
    {
      "epoch": 0.10269960164513288,
      "grad_norm": 1877.5772705078125,
      "learning_rate": 4.999888385477205e-05,
      "loss": 156.0251,
      "step": 12710
    },
    {
      "epoch": 0.10278040384941702,
      "grad_norm": 1271.5089111328125,
      "learning_rate": 4.9998816225209544e-05,
      "loss": 154.951,
      "step": 12720
    },
    {
      "epoch": 0.10286120605370115,
      "grad_norm": 1295.865966796875,
      "learning_rate": 4.9998746606605964e-05,
      "loss": 217.4705,
      "step": 12730
    },
    {
      "epoch": 0.10294200825798527,
      "grad_norm": 1123.7264404296875,
      "learning_rate": 4.999867499896684e-05,
      "loss": 162.33,
      "step": 12740
    },
    {
      "epoch": 0.10302281046226941,
      "grad_norm": 1871.114990234375,
      "learning_rate": 4.999860140229788e-05,
      "loss": 166.8098,
      "step": 12750
    },
    {
      "epoch": 0.10310361266655355,
      "grad_norm": 1002.0478515625,
      "learning_rate": 4.999852581660493e-05,
      "loss": 163.5526,
      "step": 12760
    },
    {
      "epoch": 0.10318441487083768,
      "grad_norm": 1245.7467041015625,
      "learning_rate": 4.9998448241894006e-05,
      "loss": 216.9302,
      "step": 12770
    },
    {
      "epoch": 0.1032652170751218,
      "grad_norm": 1028.039306640625,
      "learning_rate": 4.999836867817129e-05,
      "loss": 203.9537,
      "step": 12780
    },
    {
      "epoch": 0.10334601927940594,
      "grad_norm": 562.0195922851562,
      "learning_rate": 4.99982871254431e-05,
      "loss": 129.1438,
      "step": 12790
    },
    {
      "epoch": 0.10342682148369008,
      "grad_norm": 849.0570068359375,
      "learning_rate": 4.999820358371593e-05,
      "loss": 169.0486,
      "step": 12800
    },
    {
      "epoch": 0.10350762368797421,
      "grad_norm": 1529.56103515625,
      "learning_rate": 4.999811805299643e-05,
      "loss": 141.6357,
      "step": 12810
    },
    {
      "epoch": 0.10358842589225833,
      "grad_norm": 1446.593505859375,
      "learning_rate": 4.999803053329141e-05,
      "loss": 195.0215,
      "step": 12820
    },
    {
      "epoch": 0.10366922809654247,
      "grad_norm": 1069.249755859375,
      "learning_rate": 4.999794102460781e-05,
      "loss": 188.44,
      "step": 12830
    },
    {
      "epoch": 0.1037500303008266,
      "grad_norm": 1103.1290283203125,
      "learning_rate": 4.999784952695278e-05,
      "loss": 131.1704,
      "step": 12840
    },
    {
      "epoch": 0.10383083250511074,
      "grad_norm": 3719.030029296875,
      "learning_rate": 4.999775604033358e-05,
      "loss": 186.096,
      "step": 12850
    },
    {
      "epoch": 0.10391163470939488,
      "grad_norm": 934.4454345703125,
      "learning_rate": 4.9997660564757654e-05,
      "loss": 162.9967,
      "step": 12860
    },
    {
      "epoch": 0.103992436913679,
      "grad_norm": 881.9823608398438,
      "learning_rate": 4.999756310023261e-05,
      "loss": 189.1937,
      "step": 12870
    },
    {
      "epoch": 0.10407323911796314,
      "grad_norm": 4495.25634765625,
      "learning_rate": 4.9997463646766185e-05,
      "loss": 167.4466,
      "step": 12880
    },
    {
      "epoch": 0.10415404132224727,
      "grad_norm": 981.443359375,
      "learning_rate": 4.99973622043663e-05,
      "loss": 145.4228,
      "step": 12890
    },
    {
      "epoch": 0.10423484352653141,
      "grad_norm": 1170.6580810546875,
      "learning_rate": 4.9997258773041036e-05,
      "loss": 194.973,
      "step": 12900
    },
    {
      "epoch": 0.10431564573081553,
      "grad_norm": 909.8919677734375,
      "learning_rate": 4.999715335279861e-05,
      "loss": 173.4462,
      "step": 12910
    },
    {
      "epoch": 0.10439644793509967,
      "grad_norm": 756.8289184570312,
      "learning_rate": 4.9997045943647415e-05,
      "loss": 177.1729,
      "step": 12920
    },
    {
      "epoch": 0.1044772501393838,
      "grad_norm": 1074.613037109375,
      "learning_rate": 4.9996936545595986e-05,
      "loss": 152.6241,
      "step": 12930
    },
    {
      "epoch": 0.10455805234366794,
      "grad_norm": 948.0105590820312,
      "learning_rate": 4.999682515865304e-05,
      "loss": 160.605,
      "step": 12940
    },
    {
      "epoch": 0.10463885454795206,
      "grad_norm": 986.80419921875,
      "learning_rate": 4.999671178282744e-05,
      "loss": 195.0648,
      "step": 12950
    },
    {
      "epoch": 0.1047196567522362,
      "grad_norm": 1015.398193359375,
      "learning_rate": 4.99965964181282e-05,
      "loss": 152.1088,
      "step": 12960
    },
    {
      "epoch": 0.10480045895652033,
      "grad_norm": 867.7134399414062,
      "learning_rate": 4.999647906456451e-05,
      "loss": 158.7183,
      "step": 12970
    },
    {
      "epoch": 0.10488126116080447,
      "grad_norm": 1498.0318603515625,
      "learning_rate": 4.9996359722145694e-05,
      "loss": 178.2639,
      "step": 12980
    },
    {
      "epoch": 0.1049620633650886,
      "grad_norm": 1281.1707763671875,
      "learning_rate": 4.999623839088125e-05,
      "loss": 167.5088,
      "step": 12990
    },
    {
      "epoch": 0.10504286556937273,
      "grad_norm": 1084.959228515625,
      "learning_rate": 4.999611507078083e-05,
      "loss": 157.0286,
      "step": 13000
    },
    {
      "epoch": 0.10512366777365686,
      "grad_norm": 1000.6400756835938,
      "learning_rate": 4.9995989761854254e-05,
      "loss": 146.6623,
      "step": 13010
    },
    {
      "epoch": 0.105204469977941,
      "grad_norm": 1509.530029296875,
      "learning_rate": 4.99958624641115e-05,
      "loss": 177.9203,
      "step": 13020
    },
    {
      "epoch": 0.10528527218222514,
      "grad_norm": 1420.79345703125,
      "learning_rate": 4.999573317756267e-05,
      "loss": 148.3134,
      "step": 13030
    },
    {
      "epoch": 0.10536607438650926,
      "grad_norm": 1180.6531982421875,
      "learning_rate": 4.999560190221807e-05,
      "loss": 170.9963,
      "step": 13040
    },
    {
      "epoch": 0.1054468765907934,
      "grad_norm": 1092.0330810546875,
      "learning_rate": 4.999546863808815e-05,
      "loss": 133.3349,
      "step": 13050
    },
    {
      "epoch": 0.10552767879507753,
      "grad_norm": 1312.984619140625,
      "learning_rate": 4.99953333851835e-05,
      "loss": 122.4938,
      "step": 13060
    },
    {
      "epoch": 0.10560848099936167,
      "grad_norm": 1187.057861328125,
      "learning_rate": 4.999519614351488e-05,
      "loss": 137.0068,
      "step": 13070
    },
    {
      "epoch": 0.10568928320364579,
      "grad_norm": 1474.0391845703125,
      "learning_rate": 4.999505691309322e-05,
      "loss": 215.5605,
      "step": 13080
    },
    {
      "epoch": 0.10577008540792993,
      "grad_norm": 739.2951049804688,
      "learning_rate": 4.9994915693929586e-05,
      "loss": 153.0441,
      "step": 13090
    },
    {
      "epoch": 0.10585088761221406,
      "grad_norm": 576.087890625,
      "learning_rate": 4.9994772486035225e-05,
      "loss": 151.3807,
      "step": 13100
    },
    {
      "epoch": 0.1059316898164982,
      "grad_norm": 1152.282958984375,
      "learning_rate": 4.9994627289421534e-05,
      "loss": 190.3949,
      "step": 13110
    },
    {
      "epoch": 0.10601249202078232,
      "grad_norm": 2059.785888671875,
      "learning_rate": 4.999448010410005e-05,
      "loss": 140.3872,
      "step": 13120
    },
    {
      "epoch": 0.10609329422506646,
      "grad_norm": 784.2423095703125,
      "learning_rate": 4.99943309300825e-05,
      "loss": 103.543,
      "step": 13130
    },
    {
      "epoch": 0.10617409642935059,
      "grad_norm": 1098.57958984375,
      "learning_rate": 4.9994179767380746e-05,
      "loss": 150.3387,
      "step": 13140
    },
    {
      "epoch": 0.10625489863363473,
      "grad_norm": 1574.0692138671875,
      "learning_rate": 4.999402661600682e-05,
      "loss": 192.4264,
      "step": 13150
    },
    {
      "epoch": 0.10633570083791886,
      "grad_norm": 2210.006591796875,
      "learning_rate": 4.9993871475972895e-05,
      "loss": 125.3871,
      "step": 13160
    },
    {
      "epoch": 0.10641650304220299,
      "grad_norm": 1684.879150390625,
      "learning_rate": 4.999371434729132e-05,
      "loss": 205.1452,
      "step": 13170
    },
    {
      "epoch": 0.10649730524648712,
      "grad_norm": 1359.27490234375,
      "learning_rate": 4.999355522997461e-05,
      "loss": 203.7965,
      "step": 13180
    },
    {
      "epoch": 0.10657810745077126,
      "grad_norm": 916.1061401367188,
      "learning_rate": 4.999339412403541e-05,
      "loss": 162.6433,
      "step": 13190
    },
    {
      "epoch": 0.1066589096550554,
      "grad_norm": 1082.037109375,
      "learning_rate": 4.9993231029486544e-05,
      "loss": 175.1734,
      "step": 13200
    },
    {
      "epoch": 0.10673971185933952,
      "grad_norm": 1730.4796142578125,
      "learning_rate": 4.9993065946340986e-05,
      "loss": 206.4621,
      "step": 13210
    },
    {
      "epoch": 0.10682051406362365,
      "grad_norm": 1461.9183349609375,
      "learning_rate": 4.999289887461188e-05,
      "loss": 180.7627,
      "step": 13220
    },
    {
      "epoch": 0.10690131626790779,
      "grad_norm": 1014.8677368164062,
      "learning_rate": 4.999272981431251e-05,
      "loss": 179.8561,
      "step": 13230
    },
    {
      "epoch": 0.10698211847219193,
      "grad_norm": 1612.005859375,
      "learning_rate": 4.9992558765456334e-05,
      "loss": 142.1859,
      "step": 13240
    },
    {
      "epoch": 0.10706292067647605,
      "grad_norm": 958.360107421875,
      "learning_rate": 4.999238572805694e-05,
      "loss": 185.2089,
      "step": 13250
    },
    {
      "epoch": 0.10714372288076018,
      "grad_norm": 810.4140625,
      "learning_rate": 4.999221070212813e-05,
      "loss": 154.9561,
      "step": 13260
    },
    {
      "epoch": 0.10722452508504432,
      "grad_norm": 1180.060546875,
      "learning_rate": 4.999203368768381e-05,
      "loss": 152.1395,
      "step": 13270
    },
    {
      "epoch": 0.10730532728932846,
      "grad_norm": 820.518310546875,
      "learning_rate": 4.9991854684738066e-05,
      "loss": 162.0506,
      "step": 13280
    },
    {
      "epoch": 0.10738612949361259,
      "grad_norm": 654.80078125,
      "learning_rate": 4.999167369330514e-05,
      "loss": 227.117,
      "step": 13290
    },
    {
      "epoch": 0.10746693169789671,
      "grad_norm": 2003.2425537109375,
      "learning_rate": 4.9991490713399436e-05,
      "loss": 192.4266,
      "step": 13300
    },
    {
      "epoch": 0.10754773390218085,
      "grad_norm": 598.6465454101562,
      "learning_rate": 4.9991305745035514e-05,
      "loss": 153.8487,
      "step": 13310
    },
    {
      "epoch": 0.10762853610646499,
      "grad_norm": 1834.0950927734375,
      "learning_rate": 4.9991118788228084e-05,
      "loss": 177.134,
      "step": 13320
    },
    {
      "epoch": 0.10770933831074912,
      "grad_norm": 1758.453125,
      "learning_rate": 4.9990929842992026e-05,
      "loss": 170.0616,
      "step": 13330
    },
    {
      "epoch": 0.10779014051503324,
      "grad_norm": 1039.255859375,
      "learning_rate": 4.9990738909342384e-05,
      "loss": 146.2972,
      "step": 13340
    },
    {
      "epoch": 0.10787094271931738,
      "grad_norm": 1003.7833862304688,
      "learning_rate": 4.9990545987294324e-05,
      "loss": 151.9346,
      "step": 13350
    },
    {
      "epoch": 0.10795174492360152,
      "grad_norm": 5873.03857421875,
      "learning_rate": 4.999035107686322e-05,
      "loss": 144.1906,
      "step": 13360
    },
    {
      "epoch": 0.10803254712788565,
      "grad_norm": 2187.155517578125,
      "learning_rate": 4.999015417806457e-05,
      "loss": 196.4291,
      "step": 13370
    },
    {
      "epoch": 0.10811334933216978,
      "grad_norm": 1240.0516357421875,
      "learning_rate": 4.998995529091404e-05,
      "loss": 155.5472,
      "step": 13380
    },
    {
      "epoch": 0.10819415153645391,
      "grad_norm": 1135.625,
      "learning_rate": 4.998975441542745e-05,
      "loss": 197.4514,
      "step": 13390
    },
    {
      "epoch": 0.10827495374073805,
      "grad_norm": 1170.193603515625,
      "learning_rate": 4.9989551551620794e-05,
      "loss": 166.823,
      "step": 13400
    },
    {
      "epoch": 0.10835575594502218,
      "grad_norm": 1421.60791015625,
      "learning_rate": 4.998934669951021e-05,
      "loss": 154.3157,
      "step": 13410
    },
    {
      "epoch": 0.10843655814930632,
      "grad_norm": 927.9348754882812,
      "learning_rate": 4.9989139859111995e-05,
      "loss": 168.5901,
      "step": 13420
    },
    {
      "epoch": 0.10851736035359044,
      "grad_norm": 1038.8873291015625,
      "learning_rate": 4.99889310304426e-05,
      "loss": 152.3316,
      "step": 13430
    },
    {
      "epoch": 0.10859816255787458,
      "grad_norm": 947.6807250976562,
      "learning_rate": 4.9988720213518655e-05,
      "loss": 135.1435,
      "step": 13440
    },
    {
      "epoch": 0.10867896476215871,
      "grad_norm": 1577.244384765625,
      "learning_rate": 4.998850740835692e-05,
      "loss": 178.216,
      "step": 13450
    },
    {
      "epoch": 0.10875976696644285,
      "grad_norm": 885.4010620117188,
      "learning_rate": 4.998829261497433e-05,
      "loss": 203.6382,
      "step": 13460
    },
    {
      "epoch": 0.10884056917072697,
      "grad_norm": 847.8015747070312,
      "learning_rate": 4.998807583338798e-05,
      "loss": 153.3626,
      "step": 13470
    },
    {
      "epoch": 0.10892137137501111,
      "grad_norm": 1210.1986083984375,
      "learning_rate": 4.998785706361512e-05,
      "loss": 145.362,
      "step": 13480
    },
    {
      "epoch": 0.10900217357929524,
      "grad_norm": 1167.423095703125,
      "learning_rate": 4.9987636305673144e-05,
      "loss": 162.2581,
      "step": 13490
    },
    {
      "epoch": 0.10908297578357938,
      "grad_norm": 1180.3104248046875,
      "learning_rate": 4.9987413559579636e-05,
      "loss": 149.3035,
      "step": 13500
    },
    {
      "epoch": 0.1091637779878635,
      "grad_norm": 1402.9918212890625,
      "learning_rate": 4.9987188825352294e-05,
      "loss": 169.6259,
      "step": 13510
    },
    {
      "epoch": 0.10924458019214764,
      "grad_norm": 2665.6416015625,
      "learning_rate": 4.998696210300902e-05,
      "loss": 192.3846,
      "step": 13520
    },
    {
      "epoch": 0.10932538239643178,
      "grad_norm": 725.3150634765625,
      "learning_rate": 4.9986733392567845e-05,
      "loss": 197.2868,
      "step": 13530
    },
    {
      "epoch": 0.10940618460071591,
      "grad_norm": 974.5929565429688,
      "learning_rate": 4.998650269404697e-05,
      "loss": 178.5823,
      "step": 13540
    },
    {
      "epoch": 0.10948698680500005,
      "grad_norm": 1542.777587890625,
      "learning_rate": 4.998627000746475e-05,
      "loss": 200.5336,
      "step": 13550
    },
    {
      "epoch": 0.10956778900928417,
      "grad_norm": 1063.777099609375,
      "learning_rate": 4.9986035332839694e-05,
      "loss": 164.4375,
      "step": 13560
    },
    {
      "epoch": 0.1096485912135683,
      "grad_norm": 1302.0216064453125,
      "learning_rate": 4.998579867019048e-05,
      "loss": 135.7528,
      "step": 13570
    },
    {
      "epoch": 0.10972939341785244,
      "grad_norm": 807.3289794921875,
      "learning_rate": 4.998556001953593e-05,
      "loss": 156.5582,
      "step": 13580
    },
    {
      "epoch": 0.10981019562213658,
      "grad_norm": 727.23974609375,
      "learning_rate": 4.998531938089503e-05,
      "loss": 180.1156,
      "step": 13590
    },
    {
      "epoch": 0.1098909978264207,
      "grad_norm": 1151.44189453125,
      "learning_rate": 4.998507675428695e-05,
      "loss": 138.4998,
      "step": 13600
    },
    {
      "epoch": 0.10997180003070484,
      "grad_norm": 1855.6485595703125,
      "learning_rate": 4.998483213973098e-05,
      "loss": 166.6201,
      "step": 13610
    },
    {
      "epoch": 0.11005260223498897,
      "grad_norm": 1196.5108642578125,
      "learning_rate": 4.9984585537246566e-05,
      "loss": 135.4699,
      "step": 13620
    },
    {
      "epoch": 0.11013340443927311,
      "grad_norm": 1511.992919921875,
      "learning_rate": 4.998433694685335e-05,
      "loss": 136.3776,
      "step": 13630
    },
    {
      "epoch": 0.11021420664355723,
      "grad_norm": 1018.0720825195312,
      "learning_rate": 4.998408636857111e-05,
      "loss": 142.9319,
      "step": 13640
    },
    {
      "epoch": 0.11029500884784137,
      "grad_norm": 1056.8865966796875,
      "learning_rate": 4.998383380241978e-05,
      "loss": 163.6253,
      "step": 13650
    },
    {
      "epoch": 0.1103758110521255,
      "grad_norm": 568.1683959960938,
      "learning_rate": 4.9983579248419445e-05,
      "loss": 156.528,
      "step": 13660
    },
    {
      "epoch": 0.11045661325640964,
      "grad_norm": 2031.0875244140625,
      "learning_rate": 4.998332270659037e-05,
      "loss": 188.1777,
      "step": 13670
    },
    {
      "epoch": 0.11053741546069376,
      "grad_norm": 682.0467529296875,
      "learning_rate": 4.9983064176952976e-05,
      "loss": 214.7207,
      "step": 13680
    },
    {
      "epoch": 0.1106182176649779,
      "grad_norm": 1406.7552490234375,
      "learning_rate": 4.998280365952782e-05,
      "loss": 177.2549,
      "step": 13690
    },
    {
      "epoch": 0.11069901986926203,
      "grad_norm": 922.1409301757812,
      "learning_rate": 4.998254115433563e-05,
      "loss": 174.0214,
      "step": 13700
    },
    {
      "epoch": 0.11077982207354617,
      "grad_norm": 3660.348388671875,
      "learning_rate": 4.9982276661397286e-05,
      "loss": 184.1588,
      "step": 13710
    },
    {
      "epoch": 0.1108606242778303,
      "grad_norm": 1597.2998046875,
      "learning_rate": 4.998201018073385e-05,
      "loss": 152.5763,
      "step": 13720
    },
    {
      "epoch": 0.11094142648211443,
      "grad_norm": 1081.3948974609375,
      "learning_rate": 4.9981741712366515e-05,
      "loss": 179.4719,
      "step": 13730
    },
    {
      "epoch": 0.11102222868639856,
      "grad_norm": 1722.0684814453125,
      "learning_rate": 4.9981471256316645e-05,
      "loss": 166.4671,
      "step": 13740
    },
    {
      "epoch": 0.1111030308906827,
      "grad_norm": 1164.61083984375,
      "learning_rate": 4.998119881260576e-05,
      "loss": 168.6889,
      "step": 13750
    },
    {
      "epoch": 0.11118383309496684,
      "grad_norm": 798.0043334960938,
      "learning_rate": 4.998092438125552e-05,
      "loss": 120.9161,
      "step": 13760
    },
    {
      "epoch": 0.11126463529925096,
      "grad_norm": 1977.904052734375,
      "learning_rate": 4.998064796228779e-05,
      "loss": 138.4463,
      "step": 13770
    },
    {
      "epoch": 0.1113454375035351,
      "grad_norm": 784.3284301757812,
      "learning_rate": 4.998036955572453e-05,
      "loss": 160.8799,
      "step": 13780
    },
    {
      "epoch": 0.11142623970781923,
      "grad_norm": 1237.744873046875,
      "learning_rate": 4.9980089161587916e-05,
      "loss": 161.2431,
      "step": 13790
    },
    {
      "epoch": 0.11150704191210337,
      "grad_norm": 1126.8914794921875,
      "learning_rate": 4.9979806779900255e-05,
      "loss": 188.3983,
      "step": 13800
    },
    {
      "epoch": 0.11158784411638749,
      "grad_norm": 1386.9434814453125,
      "learning_rate": 4.997952241068401e-05,
      "loss": 192.1466,
      "step": 13810
    },
    {
      "epoch": 0.11166864632067162,
      "grad_norm": 680.4060668945312,
      "learning_rate": 4.99792360539618e-05,
      "loss": 120.4083,
      "step": 13820
    },
    {
      "epoch": 0.11174944852495576,
      "grad_norm": 1295.848388671875,
      "learning_rate": 4.997894770975643e-05,
      "loss": 166.7768,
      "step": 13830
    },
    {
      "epoch": 0.1118302507292399,
      "grad_norm": 1008.2403564453125,
      "learning_rate": 4.9978657378090814e-05,
      "loss": 144.6575,
      "step": 13840
    },
    {
      "epoch": 0.11191105293352403,
      "grad_norm": 1428.162353515625,
      "learning_rate": 4.997836505898807e-05,
      "loss": 154.2511,
      "step": 13850
    },
    {
      "epoch": 0.11199185513780816,
      "grad_norm": 825.681640625,
      "learning_rate": 4.997807075247146e-05,
      "loss": 121.8282,
      "step": 13860
    },
    {
      "epoch": 0.11207265734209229,
      "grad_norm": 815.547119140625,
      "learning_rate": 4.997777445856439e-05,
      "loss": 161.0432,
      "step": 13870
    },
    {
      "epoch": 0.11215345954637643,
      "grad_norm": 715.5890502929688,
      "learning_rate": 4.997747617729044e-05,
      "loss": 152.8754,
      "step": 13880
    },
    {
      "epoch": 0.11223426175066056,
      "grad_norm": 965.2642822265625,
      "learning_rate": 4.997717590867335e-05,
      "loss": 119.2692,
      "step": 13890
    },
    {
      "epoch": 0.11231506395494469,
      "grad_norm": 2111.7919921875,
      "learning_rate": 4.997687365273699e-05,
      "loss": 158.9874,
      "step": 13900
    },
    {
      "epoch": 0.11239586615922882,
      "grad_norm": 879.88037109375,
      "learning_rate": 4.9976569409505424e-05,
      "loss": 176.3235,
      "step": 13910
    },
    {
      "epoch": 0.11247666836351296,
      "grad_norm": 1380.322998046875,
      "learning_rate": 4.997626317900286e-05,
      "loss": 137.2757,
      "step": 13920
    },
    {
      "epoch": 0.1125574705677971,
      "grad_norm": 718.0297241210938,
      "learning_rate": 4.997595496125366e-05,
      "loss": 191.654,
      "step": 13930
    },
    {
      "epoch": 0.11263827277208122,
      "grad_norm": 1362.7103271484375,
      "learning_rate": 4.997564475628234e-05,
      "loss": 134.5998,
      "step": 13940
    },
    {
      "epoch": 0.11271907497636535,
      "grad_norm": 842.7426147460938,
      "learning_rate": 4.99753325641136e-05,
      "loss": 143.3749,
      "step": 13950
    },
    {
      "epoch": 0.11279987718064949,
      "grad_norm": 1636.0693359375,
      "learning_rate": 4.997501838477226e-05,
      "loss": 132.4023,
      "step": 13960
    },
    {
      "epoch": 0.11288067938493362,
      "grad_norm": 1336.4649658203125,
      "learning_rate": 4.997470221828334e-05,
      "loss": 136.2148,
      "step": 13970
    },
    {
      "epoch": 0.11296148158921776,
      "grad_norm": 906.9151000976562,
      "learning_rate": 4.997438406467197e-05,
      "loss": 175.0376,
      "step": 13980
    },
    {
      "epoch": 0.11304228379350188,
      "grad_norm": 1826.1988525390625,
      "learning_rate": 4.997406392396349e-05,
      "loss": 152.145,
      "step": 13990
    },
    {
      "epoch": 0.11312308599778602,
      "grad_norm": 1275.584228515625,
      "learning_rate": 4.997374179618335e-05,
      "loss": 190.0866,
      "step": 14000
    },
    {
      "epoch": 0.11320388820207015,
      "grad_norm": 937.593017578125,
      "learning_rate": 4.997341768135719e-05,
      "loss": 141.1656,
      "step": 14010
    },
    {
      "epoch": 0.11328469040635429,
      "grad_norm": 945.1254272460938,
      "learning_rate": 4.99730915795108e-05,
      "loss": 180.2019,
      "step": 14020
    },
    {
      "epoch": 0.11336549261063841,
      "grad_norm": 867.3080444335938,
      "learning_rate": 4.9972763490670116e-05,
      "loss": 161.3901,
      "step": 14030
    },
    {
      "epoch": 0.11344629481492255,
      "grad_norm": 1019.8523559570312,
      "learning_rate": 4.997243341486127e-05,
      "loss": 198.7109,
      "step": 14040
    },
    {
      "epoch": 0.11352709701920669,
      "grad_norm": 1486.674072265625,
      "learning_rate": 4.9972101352110476e-05,
      "loss": 223.1289,
      "step": 14050
    },
    {
      "epoch": 0.11360789922349082,
      "grad_norm": 1696.5440673828125,
      "learning_rate": 4.9971767302444204e-05,
      "loss": 145.1186,
      "step": 14060
    },
    {
      "epoch": 0.11368870142777494,
      "grad_norm": 673.3330688476562,
      "learning_rate": 4.9971431265889014e-05,
      "loss": 140.5065,
      "step": 14070
    },
    {
      "epoch": 0.11376950363205908,
      "grad_norm": 3967.926513671875,
      "learning_rate": 4.997109324247163e-05,
      "loss": 163.7622,
      "step": 14080
    },
    {
      "epoch": 0.11385030583634322,
      "grad_norm": 1310.7440185546875,
      "learning_rate": 4.997075323221897e-05,
      "loss": 165.0551,
      "step": 14090
    },
    {
      "epoch": 0.11393110804062735,
      "grad_norm": 887.0441284179688,
      "learning_rate": 4.9970411235158066e-05,
      "loss": 164.2679,
      "step": 14100
    },
    {
      "epoch": 0.11401191024491149,
      "grad_norm": 1011.033935546875,
      "learning_rate": 4.997006725131615e-05,
      "loss": 126.0271,
      "step": 14110
    },
    {
      "epoch": 0.11409271244919561,
      "grad_norm": 1387.5933837890625,
      "learning_rate": 4.996972128072057e-05,
      "loss": 159.5976,
      "step": 14120
    },
    {
      "epoch": 0.11417351465347975,
      "grad_norm": 1430.6734619140625,
      "learning_rate": 4.996937332339887e-05,
      "loss": 142.3303,
      "step": 14130
    },
    {
      "epoch": 0.11425431685776388,
      "grad_norm": 1413.59814453125,
      "learning_rate": 4.9969023379378724e-05,
      "loss": 212.5765,
      "step": 14140
    },
    {
      "epoch": 0.11433511906204802,
      "grad_norm": 981.312744140625,
      "learning_rate": 4.996867144868798e-05,
      "loss": 123.0717,
      "step": 14150
    },
    {
      "epoch": 0.11441592126633214,
      "grad_norm": 1553.310546875,
      "learning_rate": 4.996831753135464e-05,
      "loss": 187.7103,
      "step": 14160
    },
    {
      "epoch": 0.11449672347061628,
      "grad_norm": 1003.1934204101562,
      "learning_rate": 4.996796162740686e-05,
      "loss": 140.4176,
      "step": 14170
    },
    {
      "epoch": 0.11457752567490041,
      "grad_norm": 916.1061401367188,
      "learning_rate": 4.996760373687297e-05,
      "loss": 162.5518,
      "step": 14180
    },
    {
      "epoch": 0.11465832787918455,
      "grad_norm": 946.1961669921875,
      "learning_rate": 4.9967243859781426e-05,
      "loss": 156.3168,
      "step": 14190
    },
    {
      "epoch": 0.11473913008346867,
      "grad_norm": 775.9445190429688,
      "learning_rate": 4.9966881996160876e-05,
      "loss": 156.1551,
      "step": 14200
    },
    {
      "epoch": 0.11481993228775281,
      "grad_norm": 1236.88427734375,
      "learning_rate": 4.996651814604011e-05,
      "loss": 174.8899,
      "step": 14210
    },
    {
      "epoch": 0.11490073449203694,
      "grad_norm": 1576.7642822265625,
      "learning_rate": 4.9966152309448076e-05,
      "loss": 158.9026,
      "step": 14220
    },
    {
      "epoch": 0.11498153669632108,
      "grad_norm": 1118.7216796875,
      "learning_rate": 4.996578448641388e-05,
      "loss": 138.2308,
      "step": 14230
    },
    {
      "epoch": 0.1150623389006052,
      "grad_norm": 1023.9251708984375,
      "learning_rate": 4.9965414676966796e-05,
      "loss": 129.1291,
      "step": 14240
    },
    {
      "epoch": 0.11514314110488934,
      "grad_norm": 1101.9742431640625,
      "learning_rate": 4.996504288113624e-05,
      "loss": 196.3162,
      "step": 14250
    },
    {
      "epoch": 0.11522394330917347,
      "grad_norm": 796.50048828125,
      "learning_rate": 4.9964669098951786e-05,
      "loss": 147.224,
      "step": 14260
    },
    {
      "epoch": 0.11530474551345761,
      "grad_norm": 1557.001708984375,
      "learning_rate": 4.996429333044319e-05,
      "loss": 134.4548,
      "step": 14270
    },
    {
      "epoch": 0.11538554771774175,
      "grad_norm": 1414.4412841796875,
      "learning_rate": 4.996391557564035e-05,
      "loss": 153.0477,
      "step": 14280
    },
    {
      "epoch": 0.11546634992202587,
      "grad_norm": 950.5034790039062,
      "learning_rate": 4.996353583457331e-05,
      "loss": 151.1424,
      "step": 14290
    },
    {
      "epoch": 0.11554715212631,
      "grad_norm": 5219.78955078125,
      "learning_rate": 4.9963154107272295e-05,
      "loss": 194.4314,
      "step": 14300
    },
    {
      "epoch": 0.11562795433059414,
      "grad_norm": 915.6011962890625,
      "learning_rate": 4.996277039376767e-05,
      "loss": 173.4856,
      "step": 14310
    },
    {
      "epoch": 0.11570875653487828,
      "grad_norm": 817.5308837890625,
      "learning_rate": 4.996238469408997e-05,
      "loss": 108.486,
      "step": 14320
    },
    {
      "epoch": 0.1157895587391624,
      "grad_norm": 2541.830078125,
      "learning_rate": 4.996199700826988e-05,
      "loss": 220.6018,
      "step": 14330
    },
    {
      "epoch": 0.11587036094344653,
      "grad_norm": 1460.224853515625,
      "learning_rate": 4.996160733633824e-05,
      "loss": 185.1798,
      "step": 14340
    },
    {
      "epoch": 0.11595116314773067,
      "grad_norm": 1051.0694580078125,
      "learning_rate": 4.996121567832608e-05,
      "loss": 148.2247,
      "step": 14350
    },
    {
      "epoch": 0.11603196535201481,
      "grad_norm": 1205.9510498046875,
      "learning_rate": 4.9960822034264534e-05,
      "loss": 169.8678,
      "step": 14360
    },
    {
      "epoch": 0.11611276755629893,
      "grad_norm": 1407.5836181640625,
      "learning_rate": 4.996042640418494e-05,
      "loss": 158.6641,
      "step": 14370
    },
    {
      "epoch": 0.11619356976058307,
      "grad_norm": 1719.08349609375,
      "learning_rate": 4.996002878811876e-05,
      "loss": 112.0756,
      "step": 14380
    },
    {
      "epoch": 0.1162743719648672,
      "grad_norm": 1174.4710693359375,
      "learning_rate": 4.995962918609766e-05,
      "loss": 168.0695,
      "step": 14390
    },
    {
      "epoch": 0.11635517416915134,
      "grad_norm": 1517.498291015625,
      "learning_rate": 4.995922759815339e-05,
      "loss": 190.5525,
      "step": 14400
    },
    {
      "epoch": 0.11643597637343547,
      "grad_norm": 920.9449462890625,
      "learning_rate": 4.995882402431794e-05,
      "loss": 165.8069,
      "step": 14410
    },
    {
      "epoch": 0.1165167785777196,
      "grad_norm": 741.5789794921875,
      "learning_rate": 4.995841846462341e-05,
      "loss": 148.2098,
      "step": 14420
    },
    {
      "epoch": 0.11659758078200373,
      "grad_norm": 995.2381591796875,
      "learning_rate": 4.995801091910206e-05,
      "loss": 138.0731,
      "step": 14430
    },
    {
      "epoch": 0.11667838298628787,
      "grad_norm": 877.2784423828125,
      "learning_rate": 4.995760138778633e-05,
      "loss": 148.9934,
      "step": 14440
    },
    {
      "epoch": 0.116759185190572,
      "grad_norm": 1467.7303466796875,
      "learning_rate": 4.995718987070879e-05,
      "loss": 228.5873,
      "step": 14450
    },
    {
      "epoch": 0.11683998739485613,
      "grad_norm": 1128.6104736328125,
      "learning_rate": 4.99567763679022e-05,
      "loss": 149.8821,
      "step": 14460
    },
    {
      "epoch": 0.11692078959914026,
      "grad_norm": 932.7119140625,
      "learning_rate": 4.9956360879399444e-05,
      "loss": 129.9151,
      "step": 14470
    },
    {
      "epoch": 0.1170015918034244,
      "grad_norm": 1253.4259033203125,
      "learning_rate": 4.9955943405233584e-05,
      "loss": 205.2673,
      "step": 14480
    },
    {
      "epoch": 0.11708239400770853,
      "grad_norm": 2555.6826171875,
      "learning_rate": 4.995552394543784e-05,
      "loss": 143.3954,
      "step": 14490
    },
    {
      "epoch": 0.11716319621199266,
      "grad_norm": 855.253662109375,
      "learning_rate": 4.995510250004559e-05,
      "loss": 154.8423,
      "step": 14500
    },
    {
      "epoch": 0.11724399841627679,
      "grad_norm": 2018.5384521484375,
      "learning_rate": 4.9954679069090364e-05,
      "loss": 234.7133,
      "step": 14510
    },
    {
      "epoch": 0.11732480062056093,
      "grad_norm": 1564.721435546875,
      "learning_rate": 4.995425365260585e-05,
      "loss": 139.8708,
      "step": 14520
    },
    {
      "epoch": 0.11740560282484507,
      "grad_norm": 688.4728393554688,
      "learning_rate": 4.9953826250625896e-05,
      "loss": 149.8579,
      "step": 14530
    },
    {
      "epoch": 0.1174864050291292,
      "grad_norm": 2625.092529296875,
      "learning_rate": 4.995339686318451e-05,
      "loss": 158.0993,
      "step": 14540
    },
    {
      "epoch": 0.11756720723341332,
      "grad_norm": 1898.60546875,
      "learning_rate": 4.995296549031585e-05,
      "loss": 132.3351,
      "step": 14550
    },
    {
      "epoch": 0.11764800943769746,
      "grad_norm": 911.2061767578125,
      "learning_rate": 4.995253213205425e-05,
      "loss": 141.7144,
      "step": 14560
    },
    {
      "epoch": 0.1177288116419816,
      "grad_norm": 1163.8433837890625,
      "learning_rate": 4.9952096788434186e-05,
      "loss": 130.6005,
      "step": 14570
    },
    {
      "epoch": 0.11780961384626573,
      "grad_norm": 1011.4424438476562,
      "learning_rate": 4.9951659459490294e-05,
      "loss": 150.9337,
      "step": 14580
    },
    {
      "epoch": 0.11789041605054985,
      "grad_norm": 845.7263793945312,
      "learning_rate": 4.9951220145257374e-05,
      "loss": 286.6193,
      "step": 14590
    },
    {
      "epoch": 0.11797121825483399,
      "grad_norm": 1084.8226318359375,
      "learning_rate": 4.9950778845770376e-05,
      "loss": 188.6081,
      "step": 14600
    },
    {
      "epoch": 0.11805202045911813,
      "grad_norm": 1993.1123046875,
      "learning_rate": 4.9950335561064423e-05,
      "loss": 177.9271,
      "step": 14610
    },
    {
      "epoch": 0.11813282266340226,
      "grad_norm": 2498.151611328125,
      "learning_rate": 4.994989029117476e-05,
      "loss": 185.4942,
      "step": 14620
    },
    {
      "epoch": 0.11821362486768638,
      "grad_norm": 1064.155517578125,
      "learning_rate": 4.994944303613684e-05,
      "loss": 157.5091,
      "step": 14630
    },
    {
      "epoch": 0.11829442707197052,
      "grad_norm": 880.9674072265625,
      "learning_rate": 4.994899379598623e-05,
      "loss": 180.4067,
      "step": 14640
    },
    {
      "epoch": 0.11837522927625466,
      "grad_norm": 1241.9085693359375,
      "learning_rate": 4.99485425707587e-05,
      "loss": 203.0391,
      "step": 14650
    },
    {
      "epoch": 0.11845603148053879,
      "grad_norm": 803.6256713867188,
      "learning_rate": 4.994808936049013e-05,
      "loss": 183.2648,
      "step": 14660
    },
    {
      "epoch": 0.11853683368482293,
      "grad_norm": 1581.531005859375,
      "learning_rate": 4.9947634165216584e-05,
      "loss": 186.4468,
      "step": 14670
    },
    {
      "epoch": 0.11861763588910705,
      "grad_norm": 791.8948974609375,
      "learning_rate": 4.994717698497428e-05,
      "loss": 179.2291,
      "step": 14680
    },
    {
      "epoch": 0.11869843809339119,
      "grad_norm": 1237.831298828125,
      "learning_rate": 4.994671781979959e-05,
      "loss": 139.9062,
      "step": 14690
    },
    {
      "epoch": 0.11877924029767532,
      "grad_norm": 835.0073852539062,
      "learning_rate": 4.994625666972906e-05,
      "loss": 156.4611,
      "step": 14700
    },
    {
      "epoch": 0.11886004250195946,
      "grad_norm": 1219.32958984375,
      "learning_rate": 4.994579353479938e-05,
      "loss": 160.4347,
      "step": 14710
    },
    {
      "epoch": 0.11894084470624358,
      "grad_norm": 1276.5611572265625,
      "learning_rate": 4.9945328415047385e-05,
      "loss": 167.3067,
      "step": 14720
    },
    {
      "epoch": 0.11902164691052772,
      "grad_norm": 1150.7158203125,
      "learning_rate": 4.994486131051009e-05,
      "loss": 143.2454,
      "step": 14730
    },
    {
      "epoch": 0.11910244911481185,
      "grad_norm": 1355.6668701171875,
      "learning_rate": 4.994439222122468e-05,
      "loss": 156.0403,
      "step": 14740
    },
    {
      "epoch": 0.11918325131909599,
      "grad_norm": 1088.11865234375,
      "learning_rate": 4.994392114722845e-05,
      "loss": 182.5412,
      "step": 14750
    },
    {
      "epoch": 0.11926405352338011,
      "grad_norm": 896.38232421875,
      "learning_rate": 4.994344808855888e-05,
      "loss": 170.0941,
      "step": 14760
    },
    {
      "epoch": 0.11934485572766425,
      "grad_norm": 1636.67333984375,
      "learning_rate": 4.994297304525363e-05,
      "loss": 192.0554,
      "step": 14770
    },
    {
      "epoch": 0.11942565793194838,
      "grad_norm": 1122.7872314453125,
      "learning_rate": 4.994249601735049e-05,
      "loss": 139.6245,
      "step": 14780
    },
    {
      "epoch": 0.11950646013623252,
      "grad_norm": 999.4352416992188,
      "learning_rate": 4.994201700488741e-05,
      "loss": 157.1384,
      "step": 14790
    },
    {
      "epoch": 0.11958726234051664,
      "grad_norm": 1360.1336669921875,
      "learning_rate": 4.99415360079025e-05,
      "loss": 126.2787,
      "step": 14800
    },
    {
      "epoch": 0.11966806454480078,
      "grad_norm": 954.704345703125,
      "learning_rate": 4.994105302643404e-05,
      "loss": 168.1486,
      "step": 14810
    },
    {
      "epoch": 0.11974886674908491,
      "grad_norm": 817.1705932617188,
      "learning_rate": 4.994056806052046e-05,
      "loss": 133.3082,
      "step": 14820
    },
    {
      "epoch": 0.11982966895336905,
      "grad_norm": 878.759521484375,
      "learning_rate": 4.994008111020033e-05,
      "loss": 188.8091,
      "step": 14830
    },
    {
      "epoch": 0.11991047115765319,
      "grad_norm": 2776.26220703125,
      "learning_rate": 4.993959217551242e-05,
      "loss": 159.8074,
      "step": 14840
    },
    {
      "epoch": 0.11999127336193731,
      "grad_norm": 1184.76171875,
      "learning_rate": 4.993910125649561e-05,
      "loss": 192.4007,
      "step": 14850
    },
    {
      "epoch": 0.12007207556622145,
      "grad_norm": 2491.531982421875,
      "learning_rate": 4.9938608353188966e-05,
      "loss": 209.929,
      "step": 14860
    },
    {
      "epoch": 0.12015287777050558,
      "grad_norm": 1279.2767333984375,
      "learning_rate": 4.993811346563171e-05,
      "loss": 159.6929,
      "step": 14870
    },
    {
      "epoch": 0.12023367997478972,
      "grad_norm": 783.5645751953125,
      "learning_rate": 4.993761659386322e-05,
      "loss": 164.0424,
      "step": 14880
    },
    {
      "epoch": 0.12031448217907384,
      "grad_norm": 1100.378662109375,
      "learning_rate": 4.993711773792302e-05,
      "loss": 152.5349,
      "step": 14890
    },
    {
      "epoch": 0.12039528438335798,
      "grad_norm": 2542.6484375,
      "learning_rate": 4.993661689785081e-05,
      "loss": 131.4937,
      "step": 14900
    },
    {
      "epoch": 0.12047608658764211,
      "grad_norm": 1040.3431396484375,
      "learning_rate": 4.9936114073686435e-05,
      "loss": 135.133,
      "step": 14910
    },
    {
      "epoch": 0.12055688879192625,
      "grad_norm": 1289.4642333984375,
      "learning_rate": 4.9935609265469905e-05,
      "loss": 138.4034,
      "step": 14920
    },
    {
      "epoch": 0.12063769099621037,
      "grad_norm": 712.0211181640625,
      "learning_rate": 4.993510247324139e-05,
      "loss": 120.3609,
      "step": 14930
    },
    {
      "epoch": 0.1207184932004945,
      "grad_norm": 872.5120849609375,
      "learning_rate": 4.99345936970412e-05,
      "loss": 130.8901,
      "step": 14940
    },
    {
      "epoch": 0.12079929540477864,
      "grad_norm": 887.6231079101562,
      "learning_rate": 4.993408293690983e-05,
      "loss": 168.3752,
      "step": 14950
    },
    {
      "epoch": 0.12088009760906278,
      "grad_norm": 883.7645874023438,
      "learning_rate": 4.993357019288791e-05,
      "loss": 178.7722,
      "step": 14960
    },
    {
      "epoch": 0.12096089981334691,
      "grad_norm": 1578.703369140625,
      "learning_rate": 4.9933055465016245e-05,
      "loss": 226.0479,
      "step": 14970
    },
    {
      "epoch": 0.12104170201763104,
      "grad_norm": 4105.41455078125,
      "learning_rate": 4.9932538753335776e-05,
      "loss": 205.1221,
      "step": 14980
    },
    {
      "epoch": 0.12112250422191517,
      "grad_norm": 1194.9835205078125,
      "learning_rate": 4.9932020057887625e-05,
      "loss": 161.8187,
      "step": 14990
    },
    {
      "epoch": 0.12120330642619931,
      "grad_norm": 1620.5238037109375,
      "learning_rate": 4.9931499378713064e-05,
      "loss": 166.9932,
      "step": 15000
    },
    {
      "epoch": 0.12128410863048344,
      "grad_norm": 1244.3077392578125,
      "learning_rate": 4.993097671585352e-05,
      "loss": 163.8744,
      "step": 15010
    },
    {
      "epoch": 0.12136491083476757,
      "grad_norm": 639.976318359375,
      "learning_rate": 4.9930452069350566e-05,
      "loss": 191.1954,
      "step": 15020
    },
    {
      "epoch": 0.1214457130390517,
      "grad_norm": 2938.000244140625,
      "learning_rate": 4.992992543924596e-05,
      "loss": 200.1803,
      "step": 15030
    },
    {
      "epoch": 0.12152651524333584,
      "grad_norm": 2456.51953125,
      "learning_rate": 4.99293968255816e-05,
      "loss": 178.8182,
      "step": 15040
    },
    {
      "epoch": 0.12160731744761998,
      "grad_norm": 1298.4368896484375,
      "learning_rate": 4.992886622839955e-05,
      "loss": 141.6572,
      "step": 15050
    },
    {
      "epoch": 0.1216881196519041,
      "grad_norm": 930.0308227539062,
      "learning_rate": 4.9928333647742024e-05,
      "loss": 147.9233,
      "step": 15060
    },
    {
      "epoch": 0.12176892185618823,
      "grad_norm": 1263.79345703125,
      "learning_rate": 4.9927799083651385e-05,
      "loss": 130.5263,
      "step": 15070
    },
    {
      "epoch": 0.12184972406047237,
      "grad_norm": 1646.8670654296875,
      "learning_rate": 4.9927262536170183e-05,
      "loss": 167.601,
      "step": 15080
    },
    {
      "epoch": 0.1219305262647565,
      "grad_norm": 618.90234375,
      "learning_rate": 4.9926724005341095e-05,
      "loss": 168.1324,
      "step": 15090
    },
    {
      "epoch": 0.12201132846904064,
      "grad_norm": 724.7408447265625,
      "learning_rate": 4.992618349120698e-05,
      "loss": 134.3964,
      "step": 15100
    },
    {
      "epoch": 0.12209213067332476,
      "grad_norm": 1000.2115478515625,
      "learning_rate": 4.992564099381084e-05,
      "loss": 135.418,
      "step": 15110
    },
    {
      "epoch": 0.1221729328776089,
      "grad_norm": 1031.4586181640625,
      "learning_rate": 4.9925096513195846e-05,
      "loss": 151.0433,
      "step": 15120
    },
    {
      "epoch": 0.12225373508189304,
      "grad_norm": 1462.1236572265625,
      "learning_rate": 4.992455004940531e-05,
      "loss": 131.2349,
      "step": 15130
    },
    {
      "epoch": 0.12233453728617717,
      "grad_norm": 1075.0465087890625,
      "learning_rate": 4.9924001602482705e-05,
      "loss": 94.1535,
      "step": 15140
    },
    {
      "epoch": 0.1224153394904613,
      "grad_norm": 1040.51025390625,
      "learning_rate": 4.992345117247169e-05,
      "loss": 137.572,
      "step": 15150
    },
    {
      "epoch": 0.12249614169474543,
      "grad_norm": 1554.9383544921875,
      "learning_rate": 4.9922898759416046e-05,
      "loss": 188.1611,
      "step": 15160
    },
    {
      "epoch": 0.12257694389902957,
      "grad_norm": 1242.22705078125,
      "learning_rate": 4.992234436335972e-05,
      "loss": 175.7165,
      "step": 15170
    },
    {
      "epoch": 0.1226577461033137,
      "grad_norm": 1831.4871826171875,
      "learning_rate": 4.9921787984346846e-05,
      "loss": 166.0569,
      "step": 15180
    },
    {
      "epoch": 0.12273854830759783,
      "grad_norm": 775.4189453125,
      "learning_rate": 4.992122962242167e-05,
      "loss": 161.268,
      "step": 15190
    },
    {
      "epoch": 0.12281935051188196,
      "grad_norm": 1125.548095703125,
      "learning_rate": 4.992066927762862e-05,
      "loss": 149.2984,
      "step": 15200
    },
    {
      "epoch": 0.1229001527161661,
      "grad_norm": 732.4982299804688,
      "learning_rate": 4.992010695001229e-05,
      "loss": 108.5541,
      "step": 15210
    },
    {
      "epoch": 0.12298095492045023,
      "grad_norm": 980.6292724609375,
      "learning_rate": 4.9919542639617425e-05,
      "loss": 149.5483,
      "step": 15220
    },
    {
      "epoch": 0.12306175712473437,
      "grad_norm": 1520.5997314453125,
      "learning_rate": 4.991897634648891e-05,
      "loss": 135.0774,
      "step": 15230
    },
    {
      "epoch": 0.12314255932901849,
      "grad_norm": 1217.220458984375,
      "learning_rate": 4.991840807067181e-05,
      "loss": 187.3661,
      "step": 15240
    },
    {
      "epoch": 0.12322336153330263,
      "grad_norm": 821.0241088867188,
      "learning_rate": 4.991783781221134e-05,
      "loss": 190.1053,
      "step": 15250
    },
    {
      "epoch": 0.12330416373758676,
      "grad_norm": 1041.9854736328125,
      "learning_rate": 4.9917265571152875e-05,
      "loss": 157.6719,
      "step": 15260
    },
    {
      "epoch": 0.1233849659418709,
      "grad_norm": 1650.58984375,
      "learning_rate": 4.9916691347541946e-05,
      "loss": 148.1537,
      "step": 15270
    },
    {
      "epoch": 0.12346576814615502,
      "grad_norm": 1302.0968017578125,
      "learning_rate": 4.9916115141424235e-05,
      "loss": 146.3277,
      "step": 15280
    },
    {
      "epoch": 0.12354657035043916,
      "grad_norm": 1479.9271240234375,
      "learning_rate": 4.991553695284559e-05,
      "loss": 135.4141,
      "step": 15290
    },
    {
      "epoch": 0.1236273725547233,
      "grad_norm": 992.5084838867188,
      "learning_rate": 4.991495678185202e-05,
      "loss": 145.0812,
      "step": 15300
    },
    {
      "epoch": 0.12370817475900743,
      "grad_norm": 1374.5244140625,
      "learning_rate": 4.991437462848968e-05,
      "loss": 199.361,
      "step": 15310
    },
    {
      "epoch": 0.12378897696329155,
      "grad_norm": 1825.560791015625,
      "learning_rate": 4.991379049280489e-05,
      "loss": 199.0117,
      "step": 15320
    },
    {
      "epoch": 0.12386977916757569,
      "grad_norm": 1683.6776123046875,
      "learning_rate": 4.991320437484414e-05,
      "loss": 132.4031,
      "step": 15330
    },
    {
      "epoch": 0.12395058137185982,
      "grad_norm": 1264.279052734375,
      "learning_rate": 4.991261627465404e-05,
      "loss": 157.1223,
      "step": 15340
    },
    {
      "epoch": 0.12403138357614396,
      "grad_norm": 1643.26171875,
      "learning_rate": 4.99120261922814e-05,
      "loss": 170.2161,
      "step": 15350
    },
    {
      "epoch": 0.1241121857804281,
      "grad_norm": 5203.81591796875,
      "learning_rate": 4.9911434127773176e-05,
      "loss": 145.5372,
      "step": 15360
    },
    {
      "epoch": 0.12419298798471222,
      "grad_norm": 1837.298828125,
      "learning_rate": 4.991084008117646e-05,
      "loss": 135.7863,
      "step": 15370
    },
    {
      "epoch": 0.12427379018899636,
      "grad_norm": 699.8900756835938,
      "learning_rate": 4.991024405253852e-05,
      "loss": 112.2561,
      "step": 15380
    },
    {
      "epoch": 0.12435459239328049,
      "grad_norm": 970.3558959960938,
      "learning_rate": 4.990964604190679e-05,
      "loss": 133.0993,
      "step": 15390
    },
    {
      "epoch": 0.12443539459756463,
      "grad_norm": 1397.918701171875,
      "learning_rate": 4.9909046049328846e-05,
      "loss": 201.8272,
      "step": 15400
    },
    {
      "epoch": 0.12451619680184875,
      "grad_norm": 939.9736328125,
      "learning_rate": 4.990844407485242e-05,
      "loss": 127.3693,
      "step": 15410
    },
    {
      "epoch": 0.12459699900613289,
      "grad_norm": 875.4765625,
      "learning_rate": 4.9907840118525415e-05,
      "loss": 160.38,
      "step": 15420
    },
    {
      "epoch": 0.12467780121041702,
      "grad_norm": 1410.2291259765625,
      "learning_rate": 4.990723418039588e-05,
      "loss": 152.7454,
      "step": 15430
    },
    {
      "epoch": 0.12475860341470116,
      "grad_norm": 1288.0728759765625,
      "learning_rate": 4.9906626260512036e-05,
      "loss": 150.0691,
      "step": 15440
    },
    {
      "epoch": 0.12483940561898528,
      "grad_norm": 1040.0684814453125,
      "learning_rate": 4.9906016358922246e-05,
      "loss": 176.1957,
      "step": 15450
    },
    {
      "epoch": 0.12492020782326942,
      "grad_norm": 944.9694213867188,
      "learning_rate": 4.990540447567503e-05,
      "loss": 129.6401,
      "step": 15460
    },
    {
      "epoch": 0.12500101002755354,
      "grad_norm": 947.33642578125,
      "learning_rate": 4.990479061081908e-05,
      "loss": 168.0807,
      "step": 15470
    },
    {
      "epoch": 0.12508181223183767,
      "grad_norm": 1530.232666015625,
      "learning_rate": 4.9904174764403255e-05,
      "loss": 192.025,
      "step": 15480
    },
    {
      "epoch": 0.1251626144361218,
      "grad_norm": 1547.114013671875,
      "learning_rate": 4.9903556936476524e-05,
      "loss": 137.4716,
      "step": 15490
    },
    {
      "epoch": 0.12524341664040595,
      "grad_norm": 1204.82177734375,
      "learning_rate": 4.9902937127088065e-05,
      "loss": 163.7098,
      "step": 15500
    },
    {
      "epoch": 0.12532421884469008,
      "grad_norm": 1420.2991943359375,
      "learning_rate": 4.9902315336287184e-05,
      "loss": 159.6538,
      "step": 15510
    },
    {
      "epoch": 0.12540502104897422,
      "grad_norm": 1532.7950439453125,
      "learning_rate": 4.990169156412336e-05,
      "loss": 190.6898,
      "step": 15520
    },
    {
      "epoch": 0.12548582325325836,
      "grad_norm": 4929.6416015625,
      "learning_rate": 4.990106581064622e-05,
      "loss": 149.6133,
      "step": 15530
    },
    {
      "epoch": 0.1255666254575425,
      "grad_norm": 1571.04736328125,
      "learning_rate": 4.9900438075905555e-05,
      "loss": 148.4885,
      "step": 15540
    },
    {
      "epoch": 0.12564742766182663,
      "grad_norm": 724.03564453125,
      "learning_rate": 4.9899808359951314e-05,
      "loss": 172.4704,
      "step": 15550
    },
    {
      "epoch": 0.12572822986611074,
      "grad_norm": 860.8834838867188,
      "learning_rate": 4.989917666283359e-05,
      "loss": 128.226,
      "step": 15560
    },
    {
      "epoch": 0.12580903207039487,
      "grad_norm": 1097.7288818359375,
      "learning_rate": 4.9898542984602656e-05,
      "loss": 144.4699,
      "step": 15570
    },
    {
      "epoch": 0.125889834274679,
      "grad_norm": 2218.3935546875,
      "learning_rate": 4.989790732530892e-05,
      "loss": 148.3845,
      "step": 15580
    },
    {
      "epoch": 0.12597063647896314,
      "grad_norm": 548.2178955078125,
      "learning_rate": 4.989726968500297e-05,
      "loss": 122.0915,
      "step": 15590
    },
    {
      "epoch": 0.12605143868324728,
      "grad_norm": 1305.503173828125,
      "learning_rate": 4.989663006373553e-05,
      "loss": 135.4369,
      "step": 15600
    },
    {
      "epoch": 0.12613224088753142,
      "grad_norm": 1803.6846923828125,
      "learning_rate": 4.9895988461557494e-05,
      "loss": 174.447,
      "step": 15610
    },
    {
      "epoch": 0.12621304309181555,
      "grad_norm": 844.888671875,
      "learning_rate": 4.989534487851992e-05,
      "loss": 143.4764,
      "step": 15620
    },
    {
      "epoch": 0.1262938452960997,
      "grad_norm": 2333.817626953125,
      "learning_rate": 4.9894699314674006e-05,
      "loss": 192.7881,
      "step": 15630
    },
    {
      "epoch": 0.12637464750038382,
      "grad_norm": 1151.9573974609375,
      "learning_rate": 4.9894051770071113e-05,
      "loss": 106.1197,
      "step": 15640
    },
    {
      "epoch": 0.12645544970466793,
      "grad_norm": 803.3490600585938,
      "learning_rate": 4.989340224476278e-05,
      "loss": 168.2896,
      "step": 15650
    },
    {
      "epoch": 0.12653625190895207,
      "grad_norm": 1133.505126953125,
      "learning_rate": 4.9892750738800664e-05,
      "loss": 156.3058,
      "step": 15660
    },
    {
      "epoch": 0.1266170541132362,
      "grad_norm": 1112.0980224609375,
      "learning_rate": 4.989209725223662e-05,
      "loss": 99.6213,
      "step": 15670
    },
    {
      "epoch": 0.12669785631752034,
      "grad_norm": 1125.371337890625,
      "learning_rate": 4.989144178512263e-05,
      "loss": 120.4679,
      "step": 15680
    },
    {
      "epoch": 0.12677865852180448,
      "grad_norm": 1283.361083984375,
      "learning_rate": 4.9890784337510865e-05,
      "loss": 177.7757,
      "step": 15690
    },
    {
      "epoch": 0.1268594607260886,
      "grad_norm": 982.0589599609375,
      "learning_rate": 4.9890124909453615e-05,
      "loss": 154.7183,
      "step": 15700
    },
    {
      "epoch": 0.12694026293037275,
      "grad_norm": 810.0673828125,
      "learning_rate": 4.988946350100336e-05,
      "loss": 186.0556,
      "step": 15710
    },
    {
      "epoch": 0.12702106513465689,
      "grad_norm": 1088.3094482421875,
      "learning_rate": 4.988880011221272e-05,
      "loss": 139.6886,
      "step": 15720
    },
    {
      "epoch": 0.127101867338941,
      "grad_norm": 1257.6068115234375,
      "learning_rate": 4.9888134743134484e-05,
      "loss": 174.5065,
      "step": 15730
    },
    {
      "epoch": 0.12718266954322513,
      "grad_norm": 1530.7177734375,
      "learning_rate": 4.988746739382158e-05,
      "loss": 166.6874,
      "step": 15740
    },
    {
      "epoch": 0.12726347174750927,
      "grad_norm": 1231.8719482421875,
      "learning_rate": 4.988679806432712e-05,
      "loss": 170.5947,
      "step": 15750
    },
    {
      "epoch": 0.1273442739517934,
      "grad_norm": 919.1317138671875,
      "learning_rate": 4.988612675470435e-05,
      "loss": 158.5722,
      "step": 15760
    },
    {
      "epoch": 0.12742507615607754,
      "grad_norm": 691.2894897460938,
      "learning_rate": 4.988545346500668e-05,
      "loss": 170.0254,
      "step": 15770
    },
    {
      "epoch": 0.12750587836036167,
      "grad_norm": 698.147705078125,
      "learning_rate": 4.9884778195287695e-05,
      "loss": 163.0237,
      "step": 15780
    },
    {
      "epoch": 0.1275866805646458,
      "grad_norm": 1487.1177978515625,
      "learning_rate": 4.988410094560111e-05,
      "loss": 186.4031,
      "step": 15790
    },
    {
      "epoch": 0.12766748276892995,
      "grad_norm": 1065.00537109375,
      "learning_rate": 4.988342171600082e-05,
      "loss": 127.2453,
      "step": 15800
    },
    {
      "epoch": 0.12774828497321408,
      "grad_norm": 1005.9658203125,
      "learning_rate": 4.988274050654086e-05,
      "loss": 141.061,
      "step": 15810
    },
    {
      "epoch": 0.1278290871774982,
      "grad_norm": 1164.0673828125,
      "learning_rate": 4.988205731727544e-05,
      "loss": 165.4086,
      "step": 15820
    },
    {
      "epoch": 0.12790988938178233,
      "grad_norm": 7027.51513671875,
      "learning_rate": 4.988137214825891e-05,
      "loss": 220.1764,
      "step": 15830
    },
    {
      "epoch": 0.12799069158606646,
      "grad_norm": 1196.4290771484375,
      "learning_rate": 4.988068499954578e-05,
      "loss": 147.7958,
      "step": 15840
    },
    {
      "epoch": 0.1280714937903506,
      "grad_norm": 932.5397338867188,
      "learning_rate": 4.9879995871190743e-05,
      "loss": 190.3842,
      "step": 15850
    },
    {
      "epoch": 0.12815229599463474,
      "grad_norm": 864.31884765625,
      "learning_rate": 4.9879304763248615e-05,
      "loss": 149.0053,
      "step": 15860
    },
    {
      "epoch": 0.12823309819891887,
      "grad_norm": 2883.14306640625,
      "learning_rate": 4.9878611675774375e-05,
      "loss": 176.4589,
      "step": 15870
    },
    {
      "epoch": 0.128313900403203,
      "grad_norm": 950.1251831054688,
      "learning_rate": 4.9877916608823196e-05,
      "loss": 172.9278,
      "step": 15880
    },
    {
      "epoch": 0.12839470260748714,
      "grad_norm": 1128.0977783203125,
      "learning_rate": 4.9877219562450364e-05,
      "loss": 164.3597,
      "step": 15890
    },
    {
      "epoch": 0.12847550481177125,
      "grad_norm": 952.9031982421875,
      "learning_rate": 4.987652053671134e-05,
      "loss": 159.8279,
      "step": 15900
    },
    {
      "epoch": 0.1285563070160554,
      "grad_norm": 3416.225341796875,
      "learning_rate": 4.987581953166175e-05,
      "loss": 160.0613,
      "step": 15910
    },
    {
      "epoch": 0.12863710922033952,
      "grad_norm": 807.5946044921875,
      "learning_rate": 4.9875116547357356e-05,
      "loss": 150.7103,
      "step": 15920
    },
    {
      "epoch": 0.12871791142462366,
      "grad_norm": 688.88427734375,
      "learning_rate": 4.9874411583854106e-05,
      "loss": 148.7598,
      "step": 15930
    },
    {
      "epoch": 0.1287987136289078,
      "grad_norm": 914.3491821289062,
      "learning_rate": 4.987370464120808e-05,
      "loss": 124.2359,
      "step": 15940
    },
    {
      "epoch": 0.12887951583319193,
      "grad_norm": 5167.775390625,
      "learning_rate": 4.987299571947553e-05,
      "loss": 146.1354,
      "step": 15950
    },
    {
      "epoch": 0.12896031803747607,
      "grad_norm": 1595.9141845703125,
      "learning_rate": 4.9872284818712865e-05,
      "loss": 139.1633,
      "step": 15960
    },
    {
      "epoch": 0.1290411202417602,
      "grad_norm": 533.0267333984375,
      "learning_rate": 4.9871571938976645e-05,
      "loss": 167.4221,
      "step": 15970
    },
    {
      "epoch": 0.12912192244604434,
      "grad_norm": 960.4555053710938,
      "learning_rate": 4.98708570803236e-05,
      "loss": 129.6839,
      "step": 15980
    },
    {
      "epoch": 0.12920272465032845,
      "grad_norm": 603.4066772460938,
      "learning_rate": 4.9870140242810585e-05,
      "loss": 105.636,
      "step": 15990
    },
    {
      "epoch": 0.12928352685461258,
      "grad_norm": 1337.03369140625,
      "learning_rate": 4.986942142649465e-05,
      "loss": 168.6452,
      "step": 16000
    },
    {
      "epoch": 0.12936432905889672,
      "grad_norm": 2447.898193359375,
      "learning_rate": 4.9868700631432995e-05,
      "loss": 206.3615,
      "step": 16010
    },
    {
      "epoch": 0.12944513126318086,
      "grad_norm": 736.8400268554688,
      "learning_rate": 4.9867977857682965e-05,
      "loss": 145.3305,
      "step": 16020
    },
    {
      "epoch": 0.129525933467465,
      "grad_norm": 1626.4791259765625,
      "learning_rate": 4.986725310530206e-05,
      "loss": 128.715,
      "step": 16030
    },
    {
      "epoch": 0.12960673567174913,
      "grad_norm": 612.1223754882812,
      "learning_rate": 4.986652637434795e-05,
      "loss": 186.9655,
      "step": 16040
    },
    {
      "epoch": 0.12968753787603327,
      "grad_norm": 1173.4058837890625,
      "learning_rate": 4.9865797664878456e-05,
      "loss": 147.4328,
      "step": 16050
    },
    {
      "epoch": 0.1297683400803174,
      "grad_norm": 1522.8951416015625,
      "learning_rate": 4.986506697695157e-05,
      "loss": 156.7457,
      "step": 16060
    },
    {
      "epoch": 0.12984914228460154,
      "grad_norm": 520.1161499023438,
      "learning_rate": 4.986433431062541e-05,
      "loss": 114.943,
      "step": 16070
    },
    {
      "epoch": 0.12992994448888565,
      "grad_norm": 885.0980834960938,
      "learning_rate": 4.986359966595828e-05,
      "loss": 161.2493,
      "step": 16080
    },
    {
      "epoch": 0.13001074669316978,
      "grad_norm": 1161.5257568359375,
      "learning_rate": 4.9862863043008645e-05,
      "loss": 169.9536,
      "step": 16090
    },
    {
      "epoch": 0.13009154889745392,
      "grad_norm": 1123.99560546875,
      "learning_rate": 4.986212444183509e-05,
      "loss": 171.4671,
      "step": 16100
    },
    {
      "epoch": 0.13017235110173805,
      "grad_norm": 1075.032958984375,
      "learning_rate": 4.9861383862496405e-05,
      "loss": 143.3025,
      "step": 16110
    },
    {
      "epoch": 0.1302531533060222,
      "grad_norm": 2145.561279296875,
      "learning_rate": 4.9860641305051496e-05,
      "loss": 150.4522,
      "step": 16120
    },
    {
      "epoch": 0.13033395551030633,
      "grad_norm": 960.9412231445312,
      "learning_rate": 4.9859896769559454e-05,
      "loss": 125.8122,
      "step": 16130
    },
    {
      "epoch": 0.13041475771459046,
      "grad_norm": 1217.901611328125,
      "learning_rate": 4.985915025607952e-05,
      "loss": 133.196,
      "step": 16140
    },
    {
      "epoch": 0.1304955599188746,
      "grad_norm": 767.8223876953125,
      "learning_rate": 4.9858401764671095e-05,
      "loss": 150.1083,
      "step": 16150
    },
    {
      "epoch": 0.1305763621231587,
      "grad_norm": 3365.16259765625,
      "learning_rate": 4.9857651295393716e-05,
      "loss": 170.0273,
      "step": 16160
    },
    {
      "epoch": 0.13065716432744284,
      "grad_norm": 758.3485107421875,
      "learning_rate": 4.985689884830711e-05,
      "loss": 138.8482,
      "step": 16170
    },
    {
      "epoch": 0.13073796653172698,
      "grad_norm": 1844.4720458984375,
      "learning_rate": 4.985614442347114e-05,
      "loss": 192.0719,
      "step": 16180
    },
    {
      "epoch": 0.13081876873601112,
      "grad_norm": 785.1895751953125,
      "learning_rate": 4.985538802094583e-05,
      "loss": 132.7963,
      "step": 16190
    },
    {
      "epoch": 0.13089957094029525,
      "grad_norm": 1228.786376953125,
      "learning_rate": 4.985462964079137e-05,
      "loss": 150.1391,
      "step": 16200
    },
    {
      "epoch": 0.1309803731445794,
      "grad_norm": 1071.201904296875,
      "learning_rate": 4.9853869283068086e-05,
      "loss": 154.3323,
      "step": 16210
    },
    {
      "epoch": 0.13106117534886352,
      "grad_norm": 1365.28759765625,
      "learning_rate": 4.9853106947836504e-05,
      "loss": 133.5463,
      "step": 16220
    },
    {
      "epoch": 0.13114197755314766,
      "grad_norm": 1116.1046142578125,
      "learning_rate": 4.985234263515725e-05,
      "loss": 172.6865,
      "step": 16230
    },
    {
      "epoch": 0.1312227797574318,
      "grad_norm": 1164.0814208984375,
      "learning_rate": 4.985157634509115e-05,
      "loss": 140.3057,
      "step": 16240
    },
    {
      "epoch": 0.1313035819617159,
      "grad_norm": 1354.85986328125,
      "learning_rate": 4.985080807769918e-05,
      "loss": 170.2302,
      "step": 16250
    },
    {
      "epoch": 0.13138438416600004,
      "grad_norm": 1436.9825439453125,
      "learning_rate": 4.9850037833042463e-05,
      "loss": 154.777,
      "step": 16260
    },
    {
      "epoch": 0.13146518637028418,
      "grad_norm": 3008.28662109375,
      "learning_rate": 4.984926561118227e-05,
      "loss": 193.904,
      "step": 16270
    },
    {
      "epoch": 0.1315459885745683,
      "grad_norm": 2124.60693359375,
      "learning_rate": 4.984849141218007e-05,
      "loss": 181.6443,
      "step": 16280
    },
    {
      "epoch": 0.13162679077885245,
      "grad_norm": 1305.1060791015625,
      "learning_rate": 4.984771523609744e-05,
      "loss": 135.3111,
      "step": 16290
    },
    {
      "epoch": 0.13170759298313658,
      "grad_norm": 1435.72021484375,
      "learning_rate": 4.984693708299614e-05,
      "loss": 144.1642,
      "step": 16300
    },
    {
      "epoch": 0.13178839518742072,
      "grad_norm": 2482.44970703125,
      "learning_rate": 4.984615695293809e-05,
      "loss": 168.4511,
      "step": 16310
    },
    {
      "epoch": 0.13186919739170486,
      "grad_norm": 1040.33154296875,
      "learning_rate": 4.984537484598536e-05,
      "loss": 151.9577,
      "step": 16320
    },
    {
      "epoch": 0.13194999959598896,
      "grad_norm": 2147.656494140625,
      "learning_rate": 4.9844590762200185e-05,
      "loss": 156.3386,
      "step": 16330
    },
    {
      "epoch": 0.1320308018002731,
      "grad_norm": 852.7472534179688,
      "learning_rate": 4.9843804701644936e-05,
      "loss": 116.6514,
      "step": 16340
    },
    {
      "epoch": 0.13211160400455724,
      "grad_norm": 1123.03076171875,
      "learning_rate": 4.984301666438217e-05,
      "loss": 135.8654,
      "step": 16350
    },
    {
      "epoch": 0.13219240620884137,
      "grad_norm": 960.448974609375,
      "learning_rate": 4.9842226650474574e-05,
      "loss": 152.3355,
      "step": 16360
    },
    {
      "epoch": 0.1322732084131255,
      "grad_norm": 1971.15380859375,
      "learning_rate": 4.984143465998502e-05,
      "loss": 137.8897,
      "step": 16370
    },
    {
      "epoch": 0.13235401061740965,
      "grad_norm": 944.2211303710938,
      "learning_rate": 4.984064069297652e-05,
      "loss": 200.7082,
      "step": 16380
    },
    {
      "epoch": 0.13243481282169378,
      "grad_norm": 1767.5118408203125,
      "learning_rate": 4.9839844749512245e-05,
      "loss": 135.5678,
      "step": 16390
    },
    {
      "epoch": 0.13251561502597792,
      "grad_norm": 1431.9781494140625,
      "learning_rate": 4.983904682965551e-05,
      "loss": 159.7814,
      "step": 16400
    },
    {
      "epoch": 0.13259641723026205,
      "grad_norm": 722.2365112304688,
      "learning_rate": 4.9838246933469826e-05,
      "loss": 176.3924,
      "step": 16410
    },
    {
      "epoch": 0.13267721943454616,
      "grad_norm": 1308.9915771484375,
      "learning_rate": 4.9837445061018825e-05,
      "loss": 179.9182,
      "step": 16420
    },
    {
      "epoch": 0.1327580216388303,
      "grad_norm": 3178.224365234375,
      "learning_rate": 4.98366412123663e-05,
      "loss": 197.8065,
      "step": 16430
    },
    {
      "epoch": 0.13283882384311443,
      "grad_norm": 1109.3902587890625,
      "learning_rate": 4.9835835387576226e-05,
      "loss": 137.1438,
      "step": 16440
    },
    {
      "epoch": 0.13291962604739857,
      "grad_norm": 785.2881469726562,
      "learning_rate": 4.983502758671271e-05,
      "loss": 156.4888,
      "step": 16450
    },
    {
      "epoch": 0.1330004282516827,
      "grad_norm": 1051.4027099609375,
      "learning_rate": 4.9834217809840027e-05,
      "loss": 109.1863,
      "step": 16460
    },
    {
      "epoch": 0.13308123045596684,
      "grad_norm": 924.291748046875,
      "learning_rate": 4.98334060570226e-05,
      "loss": 178.5278,
      "step": 16470
    },
    {
      "epoch": 0.13316203266025098,
      "grad_norm": 842.4341430664062,
      "learning_rate": 4.983259232832503e-05,
      "loss": 166.3891,
      "step": 16480
    },
    {
      "epoch": 0.13324283486453511,
      "grad_norm": 1710.2845458984375,
      "learning_rate": 4.983177662381205e-05,
      "loss": 152.9183,
      "step": 16490
    },
    {
      "epoch": 0.13332363706881925,
      "grad_norm": 776.0635986328125,
      "learning_rate": 4.983095894354858e-05,
      "loss": 150.6734,
      "step": 16500
    },
    {
      "epoch": 0.13340443927310336,
      "grad_norm": 738.4081420898438,
      "learning_rate": 4.983013928759965e-05,
      "loss": 130.03,
      "step": 16510
    },
    {
      "epoch": 0.1334852414773875,
      "grad_norm": 1212.0933837890625,
      "learning_rate": 4.98293176560305e-05,
      "loss": 159.4638,
      "step": 16520
    },
    {
      "epoch": 0.13356604368167163,
      "grad_norm": 1721.5943603515625,
      "learning_rate": 4.982849404890649e-05,
      "loss": 146.906,
      "step": 16530
    },
    {
      "epoch": 0.13364684588595577,
      "grad_norm": 1130.90283203125,
      "learning_rate": 4.982766846629316e-05,
      "loss": 157.2213,
      "step": 16540
    },
    {
      "epoch": 0.1337276480902399,
      "grad_norm": 878.40234375,
      "learning_rate": 4.982684090825619e-05,
      "loss": 183.1443,
      "step": 16550
    },
    {
      "epoch": 0.13380845029452404,
      "grad_norm": 896.69384765625,
      "learning_rate": 4.9826011374861435e-05,
      "loss": 150.7616,
      "step": 16560
    },
    {
      "epoch": 0.13388925249880818,
      "grad_norm": 1271.7767333984375,
      "learning_rate": 4.982517986617489e-05,
      "loss": 151.5091,
      "step": 16570
    },
    {
      "epoch": 0.1339700547030923,
      "grad_norm": 786.6751098632812,
      "learning_rate": 4.982434638226271e-05,
      "loss": 140.3826,
      "step": 16580
    },
    {
      "epoch": 0.13405085690737642,
      "grad_norm": 777.1525268554688,
      "learning_rate": 4.982351092319122e-05,
      "loss": 121.1654,
      "step": 16590
    },
    {
      "epoch": 0.13413165911166056,
      "grad_norm": 1195.732666015625,
      "learning_rate": 4.982267348902688e-05,
      "loss": 129.7749,
      "step": 16600
    },
    {
      "epoch": 0.1342124613159447,
      "grad_norm": 730.3426513671875,
      "learning_rate": 4.982183407983635e-05,
      "loss": 99.6988,
      "step": 16610
    },
    {
      "epoch": 0.13429326352022883,
      "grad_norm": 1521.4287109375,
      "learning_rate": 4.982099269568639e-05,
      "loss": 134.8955,
      "step": 16620
    },
    {
      "epoch": 0.13437406572451296,
      "grad_norm": 2275.628662109375,
      "learning_rate": 4.982014933664395e-05,
      "loss": 132.2674,
      "step": 16630
    },
    {
      "epoch": 0.1344548679287971,
      "grad_norm": 1096.541259765625,
      "learning_rate": 4.981930400277614e-05,
      "loss": 169.6705,
      "step": 16640
    },
    {
      "epoch": 0.13453567013308124,
      "grad_norm": 1358.02001953125,
      "learning_rate": 4.981845669415022e-05,
      "loss": 143.6309,
      "step": 16650
    },
    {
      "epoch": 0.13461647233736537,
      "grad_norm": 655.5990600585938,
      "learning_rate": 4.9817607410833586e-05,
      "loss": 157.5329,
      "step": 16660
    },
    {
      "epoch": 0.1346972745416495,
      "grad_norm": 4219.76171875,
      "learning_rate": 4.9816756152893845e-05,
      "loss": 184.4929,
      "step": 16670
    },
    {
      "epoch": 0.13477807674593362,
      "grad_norm": 1044.2806396484375,
      "learning_rate": 4.98159029203987e-05,
      "loss": 145.4762,
      "step": 16680
    },
    {
      "epoch": 0.13485887895021775,
      "grad_norm": 786.1893920898438,
      "learning_rate": 4.9815047713416067e-05,
      "loss": 124.9497,
      "step": 16690
    },
    {
      "epoch": 0.1349396811545019,
      "grad_norm": 790.2650756835938,
      "learning_rate": 4.9814190532013955e-05,
      "loss": 136.2918,
      "step": 16700
    },
    {
      "epoch": 0.13502048335878603,
      "grad_norm": 2891.427978515625,
      "learning_rate": 4.9813331376260585e-05,
      "loss": 168.4338,
      "step": 16710
    },
    {
      "epoch": 0.13510128556307016,
      "grad_norm": 947.1802978515625,
      "learning_rate": 4.981247024622432e-05,
      "loss": 196.8272,
      "step": 16720
    },
    {
      "epoch": 0.1351820877673543,
      "grad_norm": 3245.217041015625,
      "learning_rate": 4.9811607141973674e-05,
      "loss": 161.3576,
      "step": 16730
    },
    {
      "epoch": 0.13526288997163843,
      "grad_norm": 970.99609375,
      "learning_rate": 4.981074206357731e-05,
      "loss": 118.8778,
      "step": 16740
    },
    {
      "epoch": 0.13534369217592257,
      "grad_norm": 818.5444946289062,
      "learning_rate": 4.980987501110408e-05,
      "loss": 111.06,
      "step": 16750
    },
    {
      "epoch": 0.1354244943802067,
      "grad_norm": 1678.4248046875,
      "learning_rate": 4.980900598462295e-05,
      "loss": 209.277,
      "step": 16760
    },
    {
      "epoch": 0.13550529658449081,
      "grad_norm": 782.6038208007812,
      "learning_rate": 4.980813498420306e-05,
      "loss": 148.4335,
      "step": 16770
    },
    {
      "epoch": 0.13558609878877495,
      "grad_norm": 2577.811767578125,
      "learning_rate": 4.980726200991374e-05,
      "loss": 122.7456,
      "step": 16780
    },
    {
      "epoch": 0.1356669009930591,
      "grad_norm": 1101.7432861328125,
      "learning_rate": 4.980638706182442e-05,
      "loss": 146.7527,
      "step": 16790
    },
    {
      "epoch": 0.13574770319734322,
      "grad_norm": 1047.087646484375,
      "learning_rate": 4.980551014000474e-05,
      "loss": 153.0954,
      "step": 16800
    },
    {
      "epoch": 0.13582850540162736,
      "grad_norm": 616.9617309570312,
      "learning_rate": 4.9804631244524445e-05,
      "loss": 169.0794,
      "step": 16810
    },
    {
      "epoch": 0.1359093076059115,
      "grad_norm": 3564.375,
      "learning_rate": 4.980375037545349e-05,
      "loss": 139.7387,
      "step": 16820
    },
    {
      "epoch": 0.13599010981019563,
      "grad_norm": 869.0906982421875,
      "learning_rate": 4.980286753286195e-05,
      "loss": 214.401,
      "step": 16830
    },
    {
      "epoch": 0.13607091201447977,
      "grad_norm": 1770.462646484375,
      "learning_rate": 4.980198271682007e-05,
      "loss": 164.5436,
      "step": 16840
    },
    {
      "epoch": 0.13615171421876388,
      "grad_norm": 861.6243286132812,
      "learning_rate": 4.980109592739825e-05,
      "loss": 140.7934,
      "step": 16850
    },
    {
      "epoch": 0.136232516423048,
      "grad_norm": 815.6874389648438,
      "learning_rate": 4.9800207164667044e-05,
      "loss": 148.0081,
      "step": 16860
    },
    {
      "epoch": 0.13631331862733215,
      "grad_norm": 1228.7813720703125,
      "learning_rate": 4.979931642869717e-05,
      "loss": 186.2767,
      "step": 16870
    },
    {
      "epoch": 0.13639412083161628,
      "grad_norm": 1478.2159423828125,
      "learning_rate": 4.979842371955952e-05,
      "loss": 131.0911,
      "step": 16880
    },
    {
      "epoch": 0.13647492303590042,
      "grad_norm": 2854.59228515625,
      "learning_rate": 4.979752903732509e-05,
      "loss": 165.121,
      "step": 16890
    },
    {
      "epoch": 0.13655572524018456,
      "grad_norm": 762.3851318359375,
      "learning_rate": 4.979663238206508e-05,
      "loss": 147.5221,
      "step": 16900
    },
    {
      "epoch": 0.1366365274444687,
      "grad_norm": 1374.7017822265625,
      "learning_rate": 4.979573375385083e-05,
      "loss": 170.8515,
      "step": 16910
    },
    {
      "epoch": 0.13671732964875283,
      "grad_norm": 914.9188232421875,
      "learning_rate": 4.979483315275385e-05,
      "loss": 180.3649,
      "step": 16920
    },
    {
      "epoch": 0.13679813185303696,
      "grad_norm": 787.9354248046875,
      "learning_rate": 4.979393057884578e-05,
      "loss": 137.5717,
      "step": 16930
    },
    {
      "epoch": 0.13687893405732107,
      "grad_norm": 1231.185302734375,
      "learning_rate": 4.9793026032198453e-05,
      "loss": 160.6839,
      "step": 16940
    },
    {
      "epoch": 0.1369597362616052,
      "grad_norm": 933.662841796875,
      "learning_rate": 4.9792119512883816e-05,
      "loss": 147.167,
      "step": 16950
    },
    {
      "epoch": 0.13704053846588934,
      "grad_norm": 3033.047119140625,
      "learning_rate": 4.979121102097402e-05,
      "loss": 187.1447,
      "step": 16960
    },
    {
      "epoch": 0.13712134067017348,
      "grad_norm": 4561.677734375,
      "learning_rate": 4.9790300556541334e-05,
      "loss": 158.3624,
      "step": 16970
    },
    {
      "epoch": 0.13720214287445762,
      "grad_norm": 1468.9378662109375,
      "learning_rate": 4.978938811965821e-05,
      "loss": 164.9483,
      "step": 16980
    },
    {
      "epoch": 0.13728294507874175,
      "grad_norm": 995.0773315429688,
      "learning_rate": 4.978847371039724e-05,
      "loss": 174.3868,
      "step": 16990
    },
    {
      "epoch": 0.1373637472830259,
      "grad_norm": 1231.4886474609375,
      "learning_rate": 4.978755732883118e-05,
      "loss": 183.0145,
      "step": 17000
    },
    {
      "epoch": 0.13744454948731002,
      "grad_norm": 1484.8287353515625,
      "learning_rate": 4.978663897503294e-05,
      "loss": 156.891,
      "step": 17010
    },
    {
      "epoch": 0.13752535169159413,
      "grad_norm": 1120.732421875,
      "learning_rate": 4.97857186490756e-05,
      "loss": 119.2117,
      "step": 17020
    },
    {
      "epoch": 0.13760615389587827,
      "grad_norm": 1300.58642578125,
      "learning_rate": 4.978479635103237e-05,
      "loss": 168.5496,
      "step": 17030
    },
    {
      "epoch": 0.1376869561001624,
      "grad_norm": 780.037841796875,
      "learning_rate": 4.978387208097665e-05,
      "loss": 126.0169,
      "step": 17040
    },
    {
      "epoch": 0.13776775830444654,
      "grad_norm": 1085.820068359375,
      "learning_rate": 4.978294583898196e-05,
      "loss": 143.0688,
      "step": 17050
    },
    {
      "epoch": 0.13784856050873068,
      "grad_norm": 1167.7269287109375,
      "learning_rate": 4.978201762512201e-05,
      "loss": 147.1071,
      "step": 17060
    },
    {
      "epoch": 0.1379293627130148,
      "grad_norm": 2116.8076171875,
      "learning_rate": 4.978108743947066e-05,
      "loss": 176.9654,
      "step": 17070
    },
    {
      "epoch": 0.13801016491729895,
      "grad_norm": 1101.4508056640625,
      "learning_rate": 4.97801552821019e-05,
      "loss": 147.9685,
      "step": 17080
    },
    {
      "epoch": 0.13809096712158309,
      "grad_norm": 1551.89306640625,
      "learning_rate": 4.977922115308992e-05,
      "loss": 150.4693,
      "step": 17090
    },
    {
      "epoch": 0.13817176932586722,
      "grad_norm": 1168.867431640625,
      "learning_rate": 4.977828505250903e-05,
      "loss": 161.2141,
      "step": 17100
    },
    {
      "epoch": 0.13825257153015133,
      "grad_norm": 759.61962890625,
      "learning_rate": 4.977734698043371e-05,
      "loss": 150.161,
      "step": 17110
    },
    {
      "epoch": 0.13833337373443547,
      "grad_norm": 757.785400390625,
      "learning_rate": 4.977640693693862e-05,
      "loss": 135.1811,
      "step": 17120
    },
    {
      "epoch": 0.1384141759387196,
      "grad_norm": 2548.685302734375,
      "learning_rate": 4.9775464922098524e-05,
      "loss": 188.9629,
      "step": 17130
    },
    {
      "epoch": 0.13849497814300374,
      "grad_norm": 911.2066650390625,
      "learning_rate": 4.977452093598839e-05,
      "loss": 138.8588,
      "step": 17140
    },
    {
      "epoch": 0.13857578034728787,
      "grad_norm": 1549.0982666015625,
      "learning_rate": 4.977357497868334e-05,
      "loss": 162.1509,
      "step": 17150
    },
    {
      "epoch": 0.138656582551572,
      "grad_norm": 1119.6712646484375,
      "learning_rate": 4.9772627050258604e-05,
      "loss": 181.102,
      "step": 17160
    },
    {
      "epoch": 0.13873738475585615,
      "grad_norm": 817.0682983398438,
      "learning_rate": 4.977167715078963e-05,
      "loss": 167.9592,
      "step": 17170
    },
    {
      "epoch": 0.13881818696014028,
      "grad_norm": 734.8985595703125,
      "learning_rate": 4.977072528035199e-05,
      "loss": 134.1706,
      "step": 17180
    },
    {
      "epoch": 0.13889898916442442,
      "grad_norm": 927.8806762695312,
      "learning_rate": 4.976977143902143e-05,
      "loss": 150.6093,
      "step": 17190
    },
    {
      "epoch": 0.13897979136870853,
      "grad_norm": 1223.3267822265625,
      "learning_rate": 4.9768815626873836e-05,
      "loss": 140.2831,
      "step": 17200
    },
    {
      "epoch": 0.13906059357299266,
      "grad_norm": 966.7178955078125,
      "learning_rate": 4.9767857843985245e-05,
      "loss": 137.6861,
      "step": 17210
    },
    {
      "epoch": 0.1391413957772768,
      "grad_norm": 1357.1478271484375,
      "learning_rate": 4.976689809043188e-05,
      "loss": 188.3903,
      "step": 17220
    },
    {
      "epoch": 0.13922219798156094,
      "grad_norm": 1198.27001953125,
      "learning_rate": 4.97659363662901e-05,
      "loss": 112.1936,
      "step": 17230
    },
    {
      "epoch": 0.13930300018584507,
      "grad_norm": 1542.5234375,
      "learning_rate": 4.976497267163642e-05,
      "loss": 178.1322,
      "step": 17240
    },
    {
      "epoch": 0.1393838023901292,
      "grad_norm": 574.7208862304688,
      "learning_rate": 4.9764007006547516e-05,
      "loss": 130.098,
      "step": 17250
    },
    {
      "epoch": 0.13946460459441334,
      "grad_norm": 1796.7242431640625,
      "learning_rate": 4.976303937110024e-05,
      "loss": 145.7055,
      "step": 17260
    },
    {
      "epoch": 0.13954540679869748,
      "grad_norm": 958.0619506835938,
      "learning_rate": 4.9762069765371556e-05,
      "loss": 134.8362,
      "step": 17270
    },
    {
      "epoch": 0.1396262090029816,
      "grad_norm": 940.2138671875,
      "learning_rate": 4.976109818943863e-05,
      "loss": 182.4838,
      "step": 17280
    },
    {
      "epoch": 0.13970701120726572,
      "grad_norm": 547.7529907226562,
      "learning_rate": 4.976012464337876e-05,
      "loss": 112.5585,
      "step": 17290
    },
    {
      "epoch": 0.13978781341154986,
      "grad_norm": 1002.335693359375,
      "learning_rate": 4.97591491272694e-05,
      "loss": 158.1674,
      "step": 17300
    },
    {
      "epoch": 0.139868615615834,
      "grad_norm": 1832.1202392578125,
      "learning_rate": 4.9758171641188174e-05,
      "loss": 159.6637,
      "step": 17310
    },
    {
      "epoch": 0.13994941782011813,
      "grad_norm": 2267.85546875,
      "learning_rate": 4.975719218521285e-05,
      "loss": 121.5045,
      "step": 17320
    },
    {
      "epoch": 0.14003022002440227,
      "grad_norm": 1080.1688232421875,
      "learning_rate": 4.975621075942137e-05,
      "loss": 145.8474,
      "step": 17330
    },
    {
      "epoch": 0.1401110222286864,
      "grad_norm": 5002.1943359375,
      "learning_rate": 4.975522736389182e-05,
      "loss": 162.2757,
      "step": 17340
    },
    {
      "epoch": 0.14019182443297054,
      "grad_norm": 1529.05810546875,
      "learning_rate": 4.975424199870244e-05,
      "loss": 204.3265,
      "step": 17350
    },
    {
      "epoch": 0.14027262663725468,
      "grad_norm": 827.3128662109375,
      "learning_rate": 4.975325466393163e-05,
      "loss": 171.37,
      "step": 17360
    },
    {
      "epoch": 0.14035342884153879,
      "grad_norm": 802.5576782226562,
      "learning_rate": 4.975226535965795e-05,
      "loss": 172.4737,
      "step": 17370
    },
    {
      "epoch": 0.14043423104582292,
      "grad_norm": 954.416015625,
      "learning_rate": 4.9751274085960097e-05,
      "loss": 198.0529,
      "step": 17380
    },
    {
      "epoch": 0.14051503325010706,
      "grad_norm": 978.6486206054688,
      "learning_rate": 4.975028084291697e-05,
      "loss": 137.7169,
      "step": 17390
    },
    {
      "epoch": 0.1405958354543912,
      "grad_norm": 785.1177368164062,
      "learning_rate": 4.9749285630607587e-05,
      "loss": 141.8211,
      "step": 17400
    },
    {
      "epoch": 0.14067663765867533,
      "grad_norm": 1156.6881103515625,
      "learning_rate": 4.9748288449111126e-05,
      "loss": 147.5348,
      "step": 17410
    },
    {
      "epoch": 0.14075743986295947,
      "grad_norm": 1045.3648681640625,
      "learning_rate": 4.974728929850694e-05,
      "loss": 161.647,
      "step": 17420
    },
    {
      "epoch": 0.1408382420672436,
      "grad_norm": 1343.872802734375,
      "learning_rate": 4.974628817887451e-05,
      "loss": 145.1868,
      "step": 17430
    },
    {
      "epoch": 0.14091904427152774,
      "grad_norm": 1589.7122802734375,
      "learning_rate": 4.97452850902935e-05,
      "loss": 118.9889,
      "step": 17440
    },
    {
      "epoch": 0.14099984647581185,
      "grad_norm": 706.4135131835938,
      "learning_rate": 4.9744280032843726e-05,
      "loss": 157.0399,
      "step": 17450
    },
    {
      "epoch": 0.14108064868009598,
      "grad_norm": 1923.8131103515625,
      "learning_rate": 4.974327300660515e-05,
      "loss": 152.5646,
      "step": 17460
    },
    {
      "epoch": 0.14116145088438012,
      "grad_norm": 1136.072021484375,
      "learning_rate": 4.974226401165789e-05,
      "loss": 163.3207,
      "step": 17470
    },
    {
      "epoch": 0.14124225308866425,
      "grad_norm": 1105.5438232421875,
      "learning_rate": 4.974125304808224e-05,
      "loss": 150.4518,
      "step": 17480
    },
    {
      "epoch": 0.1413230552929484,
      "grad_norm": 1104.9368896484375,
      "learning_rate": 4.974024011595864e-05,
      "loss": 163.8363,
      "step": 17490
    },
    {
      "epoch": 0.14140385749723253,
      "grad_norm": 855.8195190429688,
      "learning_rate": 4.973922521536766e-05,
      "loss": 242.4139,
      "step": 17500
    },
    {
      "epoch": 0.14148465970151666,
      "grad_norm": 1221.6298828125,
      "learning_rate": 4.973820834639008e-05,
      "loss": 148.5189,
      "step": 17510
    },
    {
      "epoch": 0.1415654619058008,
      "grad_norm": 1131.8109130859375,
      "learning_rate": 4.973718950910679e-05,
      "loss": 156.5813,
      "step": 17520
    },
    {
      "epoch": 0.14164626411008494,
      "grad_norm": 1203.230712890625,
      "learning_rate": 4.973616870359886e-05,
      "loss": 149.7097,
      "step": 17530
    },
    {
      "epoch": 0.14172706631436904,
      "grad_norm": 1183.5244140625,
      "learning_rate": 4.9735145929947506e-05,
      "loss": 159.902,
      "step": 17540
    },
    {
      "epoch": 0.14180786851865318,
      "grad_norm": 1980.1556396484375,
      "learning_rate": 4.973412118823412e-05,
      "loss": 138.3822,
      "step": 17550
    },
    {
      "epoch": 0.14188867072293732,
      "grad_norm": 2065.0751953125,
      "learning_rate": 4.973309447854021e-05,
      "loss": 187.6551,
      "step": 17560
    },
    {
      "epoch": 0.14196947292722145,
      "grad_norm": 466.6748962402344,
      "learning_rate": 4.973206580094749e-05,
      "loss": 164.5834,
      "step": 17570
    },
    {
      "epoch": 0.1420502751315056,
      "grad_norm": 1023.6004028320312,
      "learning_rate": 4.9731035155537805e-05,
      "loss": 142.399,
      "step": 17580
    },
    {
      "epoch": 0.14213107733578972,
      "grad_norm": 1570.9815673828125,
      "learning_rate": 4.973000254239314e-05,
      "loss": 150.091,
      "step": 17590
    },
    {
      "epoch": 0.14221187954007386,
      "grad_norm": 886.7949829101562,
      "learning_rate": 4.972896796159568e-05,
      "loss": 215.1895,
      "step": 17600
    },
    {
      "epoch": 0.142292681744358,
      "grad_norm": 2179.057373046875,
      "learning_rate": 4.972793141322773e-05,
      "loss": 134.2585,
      "step": 17610
    },
    {
      "epoch": 0.14237348394864213,
      "grad_norm": 1217.0042724609375,
      "learning_rate": 4.9726892897371754e-05,
      "loss": 172.0116,
      "step": 17620
    },
    {
      "epoch": 0.14245428615292624,
      "grad_norm": 1240.138671875,
      "learning_rate": 4.9725852414110396e-05,
      "loss": 150.7606,
      "step": 17630
    },
    {
      "epoch": 0.14253508835721038,
      "grad_norm": 1866.2113037109375,
      "learning_rate": 4.972480996352644e-05,
      "loss": 169.5855,
      "step": 17640
    },
    {
      "epoch": 0.1426158905614945,
      "grad_norm": 1343.4752197265625,
      "learning_rate": 4.972376554570282e-05,
      "loss": 138.5565,
      "step": 17650
    },
    {
      "epoch": 0.14269669276577865,
      "grad_norm": 3173.331787109375,
      "learning_rate": 4.972271916072264e-05,
      "loss": 161.6781,
      "step": 17660
    },
    {
      "epoch": 0.14277749497006278,
      "grad_norm": 1009.836669921875,
      "learning_rate": 4.972167080866917e-05,
      "loss": 167.5559,
      "step": 17670
    },
    {
      "epoch": 0.14285829717434692,
      "grad_norm": 1085.6519775390625,
      "learning_rate": 4.9720620489625804e-05,
      "loss": 163.0791,
      "step": 17680
    },
    {
      "epoch": 0.14293909937863106,
      "grad_norm": 533.2945556640625,
      "learning_rate": 4.971956820367612e-05,
      "loss": 136.0262,
      "step": 17690
    },
    {
      "epoch": 0.1430199015829152,
      "grad_norm": 1245.5428466796875,
      "learning_rate": 4.971851395090384e-05,
      "loss": 175.3457,
      "step": 17700
    },
    {
      "epoch": 0.1431007037871993,
      "grad_norm": 1112.46630859375,
      "learning_rate": 4.9717457731392854e-05,
      "loss": 153.0563,
      "step": 17710
    },
    {
      "epoch": 0.14318150599148344,
      "grad_norm": 1563.6947021484375,
      "learning_rate": 4.971639954522719e-05,
      "loss": 170.2951,
      "step": 17720
    },
    {
      "epoch": 0.14326230819576757,
      "grad_norm": 4072.32568359375,
      "learning_rate": 4.971533939249105e-05,
      "loss": 154.5233,
      "step": 17730
    },
    {
      "epoch": 0.1433431104000517,
      "grad_norm": 995.8030395507812,
      "learning_rate": 4.971427727326877e-05,
      "loss": 135.2525,
      "step": 17740
    },
    {
      "epoch": 0.14342391260433585,
      "grad_norm": 1153.982666015625,
      "learning_rate": 4.971321318764488e-05,
      "loss": 145.9184,
      "step": 17750
    },
    {
      "epoch": 0.14350471480861998,
      "grad_norm": 1135.168212890625,
      "learning_rate": 4.971214713570403e-05,
      "loss": 106.7287,
      "step": 17760
    },
    {
      "epoch": 0.14358551701290412,
      "grad_norm": 2647.79736328125,
      "learning_rate": 4.9711079117531054e-05,
      "loss": 140.8377,
      "step": 17770
    },
    {
      "epoch": 0.14366631921718825,
      "grad_norm": 720.4559936523438,
      "learning_rate": 4.9710009133210915e-05,
      "loss": 223.2201,
      "step": 17780
    },
    {
      "epoch": 0.1437471214214724,
      "grad_norm": 877.2671508789062,
      "learning_rate": 4.970893718282876e-05,
      "loss": 126.1438,
      "step": 17790
    },
    {
      "epoch": 0.1438279236257565,
      "grad_norm": 1463.1614990234375,
      "learning_rate": 4.970786326646987e-05,
      "loss": 158.2247,
      "step": 17800
    },
    {
      "epoch": 0.14390872583004063,
      "grad_norm": 946.1857299804688,
      "learning_rate": 4.970678738421969e-05,
      "loss": 124.8285,
      "step": 17810
    },
    {
      "epoch": 0.14398952803432477,
      "grad_norm": 1228.1309814453125,
      "learning_rate": 4.9705709536163824e-05,
      "loss": 133.9969,
      "step": 17820
    },
    {
      "epoch": 0.1440703302386089,
      "grad_norm": 1009.2578735351562,
      "learning_rate": 4.9704629722388035e-05,
      "loss": 129.6919,
      "step": 17830
    },
    {
      "epoch": 0.14415113244289304,
      "grad_norm": 1004.6964111328125,
      "learning_rate": 4.9703547942978244e-05,
      "loss": 162.5422,
      "step": 17840
    },
    {
      "epoch": 0.14423193464717718,
      "grad_norm": 1071.4462890625,
      "learning_rate": 4.9702464198020517e-05,
      "loss": 118.5496,
      "step": 17850
    },
    {
      "epoch": 0.14431273685146132,
      "grad_norm": 733.4691162109375,
      "learning_rate": 4.9701378487601074e-05,
      "loss": 183.4323,
      "step": 17860
    },
    {
      "epoch": 0.14439353905574545,
      "grad_norm": 1384.6099853515625,
      "learning_rate": 4.970029081180632e-05,
      "loss": 126.8512,
      "step": 17870
    },
    {
      "epoch": 0.1444743412600296,
      "grad_norm": 1482.7987060546875,
      "learning_rate": 4.969920117072277e-05,
      "loss": 193.9891,
      "step": 17880
    },
    {
      "epoch": 0.1445551434643137,
      "grad_norm": 1112.25390625,
      "learning_rate": 4.969810956443715e-05,
      "loss": 163.0959,
      "step": 17890
    },
    {
      "epoch": 0.14463594566859783,
      "grad_norm": 1225.2054443359375,
      "learning_rate": 4.96970159930363e-05,
      "loss": 185.9541,
      "step": 17900
    },
    {
      "epoch": 0.14471674787288197,
      "grad_norm": 542.3731689453125,
      "learning_rate": 4.9695920456607226e-05,
      "loss": 171.242,
      "step": 17910
    },
    {
      "epoch": 0.1447975500771661,
      "grad_norm": 713.3396606445312,
      "learning_rate": 4.96948229552371e-05,
      "loss": 129.3645,
      "step": 17920
    },
    {
      "epoch": 0.14487835228145024,
      "grad_norm": 1782.2279052734375,
      "learning_rate": 4.9693723489013253e-05,
      "loss": 149.2307,
      "step": 17930
    },
    {
      "epoch": 0.14495915448573438,
      "grad_norm": 791.181640625,
      "learning_rate": 4.969262205802315e-05,
      "loss": 186.5776,
      "step": 17940
    },
    {
      "epoch": 0.1450399566900185,
      "grad_norm": 1428.761962890625,
      "learning_rate": 4.9691518662354434e-05,
      "loss": 142.7595,
      "step": 17950
    },
    {
      "epoch": 0.14512075889430265,
      "grad_norm": 897.8218994140625,
      "learning_rate": 4.96904133020949e-05,
      "loss": 127.4383,
      "step": 17960
    },
    {
      "epoch": 0.14520156109858676,
      "grad_norm": 1025.088623046875,
      "learning_rate": 4.968930597733249e-05,
      "loss": 184.0482,
      "step": 17970
    },
    {
      "epoch": 0.1452823633028709,
      "grad_norm": 1242.26220703125,
      "learning_rate": 4.968819668815532e-05,
      "loss": 102.1975,
      "step": 17980
    },
    {
      "epoch": 0.14536316550715503,
      "grad_norm": 1774.1246337890625,
      "learning_rate": 4.9687085434651636e-05,
      "loss": 133.0901,
      "step": 17990
    },
    {
      "epoch": 0.14544396771143916,
      "grad_norm": 1936.0460205078125,
      "learning_rate": 4.968597221690986e-05,
      "loss": 196.2127,
      "step": 18000
    },
    {
      "epoch": 0.1455247699157233,
      "grad_norm": 1869.914306640625,
      "learning_rate": 4.968485703501857e-05,
      "loss": 112.3861,
      "step": 18010
    },
    {
      "epoch": 0.14560557212000744,
      "grad_norm": 792.3851318359375,
      "learning_rate": 4.9683739889066497e-05,
      "loss": 169.0835,
      "step": 18020
    },
    {
      "epoch": 0.14568637432429157,
      "grad_norm": 1420.177734375,
      "learning_rate": 4.968262077914252e-05,
      "loss": 164.7283,
      "step": 18030
    },
    {
      "epoch": 0.1457671765285757,
      "grad_norm": 850.1755981445312,
      "learning_rate": 4.9681499705335685e-05,
      "loss": 136.3054,
      "step": 18040
    },
    {
      "epoch": 0.14584797873285985,
      "grad_norm": 729.207763671875,
      "learning_rate": 4.96803766677352e-05,
      "loss": 172.2848,
      "step": 18050
    },
    {
      "epoch": 0.14592878093714395,
      "grad_norm": 1386.603271484375,
      "learning_rate": 4.96792516664304e-05,
      "loss": 182.5913,
      "step": 18060
    },
    {
      "epoch": 0.1460095831414281,
      "grad_norm": 900.0386962890625,
      "learning_rate": 4.967812470151082e-05,
      "loss": 165.9354,
      "step": 18070
    },
    {
      "epoch": 0.14609038534571223,
      "grad_norm": 940.3046875,
      "learning_rate": 4.9676995773066105e-05,
      "loss": 112.6717,
      "step": 18080
    },
    {
      "epoch": 0.14617118754999636,
      "grad_norm": 4443.7900390625,
      "learning_rate": 4.967586488118609e-05,
      "loss": 161.5711,
      "step": 18090
    },
    {
      "epoch": 0.1462519897542805,
      "grad_norm": 955.7782592773438,
      "learning_rate": 4.9674732025960755e-05,
      "loss": 194.389,
      "step": 18100
    },
    {
      "epoch": 0.14633279195856463,
      "grad_norm": 2043.8023681640625,
      "learning_rate": 4.9673597207480236e-05,
      "loss": 147.0529,
      "step": 18110
    },
    {
      "epoch": 0.14641359416284877,
      "grad_norm": 2997.87890625,
      "learning_rate": 4.967246042583482e-05,
      "loss": 146.364,
      "step": 18120
    },
    {
      "epoch": 0.1464943963671329,
      "grad_norm": 1237.5732421875,
      "learning_rate": 4.967132168111496e-05,
      "loss": 132.0931,
      "step": 18130
    },
    {
      "epoch": 0.14657519857141701,
      "grad_norm": 1432.5120849609375,
      "learning_rate": 4.967018097341126e-05,
      "loss": 213.1167,
      "step": 18140
    },
    {
      "epoch": 0.14665600077570115,
      "grad_norm": 1035.17724609375,
      "learning_rate": 4.966903830281449e-05,
      "loss": 133.9097,
      "step": 18150
    },
    {
      "epoch": 0.1467368029799853,
      "grad_norm": 879.0906982421875,
      "learning_rate": 4.9667893669415546e-05,
      "loss": 161.3391,
      "step": 18160
    },
    {
      "epoch": 0.14681760518426942,
      "grad_norm": 1065.3148193359375,
      "learning_rate": 4.966674707330551e-05,
      "loss": 147.773,
      "step": 18170
    },
    {
      "epoch": 0.14689840738855356,
      "grad_norm": 993.1563720703125,
      "learning_rate": 4.966559851457562e-05,
      "loss": 144.7801,
      "step": 18180
    },
    {
      "epoch": 0.1469792095928377,
      "grad_norm": 551.8864135742188,
      "learning_rate": 4.966444799331726e-05,
      "loss": 142.696,
      "step": 18190
    },
    {
      "epoch": 0.14706001179712183,
      "grad_norm": 641.4093017578125,
      "learning_rate": 4.966329550962196e-05,
      "loss": 134.7732,
      "step": 18200
    },
    {
      "epoch": 0.14714081400140597,
      "grad_norm": 853.4235229492188,
      "learning_rate": 4.9662141063581436e-05,
      "loss": 142.1104,
      "step": 18210
    },
    {
      "epoch": 0.1472216162056901,
      "grad_norm": 1158.81005859375,
      "learning_rate": 4.9660984655287525e-05,
      "loss": 161.9792,
      "step": 18220
    },
    {
      "epoch": 0.1473024184099742,
      "grad_norm": 1042.7049560546875,
      "learning_rate": 4.965982628483224e-05,
      "loss": 155.6552,
      "step": 18230
    },
    {
      "epoch": 0.14738322061425835,
      "grad_norm": 746.765380859375,
      "learning_rate": 4.965866595230776e-05,
      "loss": 121.6951,
      "step": 18240
    },
    {
      "epoch": 0.14746402281854248,
      "grad_norm": 687.2789916992188,
      "learning_rate": 4.9657503657806395e-05,
      "loss": 113.0879,
      "step": 18250
    },
    {
      "epoch": 0.14754482502282662,
      "grad_norm": 1133.392822265625,
      "learning_rate": 4.9656339401420624e-05,
      "loss": 166.3725,
      "step": 18260
    },
    {
      "epoch": 0.14762562722711076,
      "grad_norm": 733.4806518554688,
      "learning_rate": 4.965517318324308e-05,
      "loss": 139.1904,
      "step": 18270
    },
    {
      "epoch": 0.1477064294313949,
      "grad_norm": 1013.0859375,
      "learning_rate": 4.9654005003366566e-05,
      "loss": 136.2502,
      "step": 18280
    },
    {
      "epoch": 0.14778723163567903,
      "grad_norm": 708.066162109375,
      "learning_rate": 4.965283486188401e-05,
      "loss": 193.8193,
      "step": 18290
    },
    {
      "epoch": 0.14786803383996316,
      "grad_norm": 1149.5897216796875,
      "learning_rate": 4.965166275888854e-05,
      "loss": 131.3347,
      "step": 18300
    },
    {
      "epoch": 0.1479488360442473,
      "grad_norm": 728.546142578125,
      "learning_rate": 4.965048869447339e-05,
      "loss": 170.7203,
      "step": 18310
    },
    {
      "epoch": 0.1480296382485314,
      "grad_norm": 1086.8751220703125,
      "learning_rate": 4.964931266873198e-05,
      "loss": 140.5396,
      "step": 18320
    },
    {
      "epoch": 0.14811044045281554,
      "grad_norm": 2335.054931640625,
      "learning_rate": 4.96481346817579e-05,
      "loss": 190.4257,
      "step": 18330
    },
    {
      "epoch": 0.14819124265709968,
      "grad_norm": 889.5890502929688,
      "learning_rate": 4.9646954733644856e-05,
      "loss": 143.3664,
      "step": 18340
    },
    {
      "epoch": 0.14827204486138382,
      "grad_norm": 638.7859497070312,
      "learning_rate": 4.9645772824486734e-05,
      "loss": 130.9303,
      "step": 18350
    },
    {
      "epoch": 0.14835284706566795,
      "grad_norm": 429.1936340332031,
      "learning_rate": 4.964458895437759e-05,
      "loss": 194.556,
      "step": 18360
    },
    {
      "epoch": 0.1484336492699521,
      "grad_norm": 625.8038330078125,
      "learning_rate": 4.96434031234116e-05,
      "loss": 137.4827,
      "step": 18370
    },
    {
      "epoch": 0.14851445147423623,
      "grad_norm": 1951.56005859375,
      "learning_rate": 4.964221533168312e-05,
      "loss": 177.7926,
      "step": 18380
    },
    {
      "epoch": 0.14859525367852036,
      "grad_norm": 2383.371337890625,
      "learning_rate": 4.9641025579286656e-05,
      "loss": 142.2735,
      "step": 18390
    },
    {
      "epoch": 0.14867605588280447,
      "grad_norm": 1063.8594970703125,
      "learning_rate": 4.9639833866316874e-05,
      "loss": 138.8885,
      "step": 18400
    },
    {
      "epoch": 0.1487568580870886,
      "grad_norm": 1529.0601806640625,
      "learning_rate": 4.963864019286859e-05,
      "loss": 140.622,
      "step": 18410
    },
    {
      "epoch": 0.14883766029137274,
      "grad_norm": 1062.4075927734375,
      "learning_rate": 4.963744455903679e-05,
      "loss": 167.9912,
      "step": 18420
    },
    {
      "epoch": 0.14891846249565688,
      "grad_norm": 590.7759399414062,
      "learning_rate": 4.963624696491659e-05,
      "loss": 130.7837,
      "step": 18430
    },
    {
      "epoch": 0.14899926469994101,
      "grad_norm": 991.8165283203125,
      "learning_rate": 4.963504741060329e-05,
      "loss": 156.975,
      "step": 18440
    },
    {
      "epoch": 0.14908006690422515,
      "grad_norm": 1038.615966796875,
      "learning_rate": 4.963384589619233e-05,
      "loss": 135.4892,
      "step": 18450
    },
    {
      "epoch": 0.1491608691085093,
      "grad_norm": 738.695556640625,
      "learning_rate": 4.9632642421779295e-05,
      "loss": 129.8741,
      "step": 18460
    },
    {
      "epoch": 0.14924167131279342,
      "grad_norm": 1388.894287109375,
      "learning_rate": 4.9631436987459964e-05,
      "loss": 140.8633,
      "step": 18470
    },
    {
      "epoch": 0.14932247351707756,
      "grad_norm": 1890.2734375,
      "learning_rate": 4.9630229593330226e-05,
      "loss": 195.7736,
      "step": 18480
    },
    {
      "epoch": 0.14940327572136167,
      "grad_norm": 1176.2431640625,
      "learning_rate": 4.9629020239486155e-05,
      "loss": 176.4671,
      "step": 18490
    },
    {
      "epoch": 0.1494840779256458,
      "grad_norm": 1625.3804931640625,
      "learning_rate": 4.962780892602398e-05,
      "loss": 182.2511,
      "step": 18500
    },
    {
      "epoch": 0.14956488012992994,
      "grad_norm": 1016.1923828125,
      "learning_rate": 4.962659565304008e-05,
      "loss": 139.6862,
      "step": 18510
    },
    {
      "epoch": 0.14964568233421408,
      "grad_norm": 1304.57373046875,
      "learning_rate": 4.962538042063097e-05,
      "loss": 133.6651,
      "step": 18520
    },
    {
      "epoch": 0.1497264845384982,
      "grad_norm": 2883.068359375,
      "learning_rate": 4.962416322889337e-05,
      "loss": 155.5502,
      "step": 18530
    },
    {
      "epoch": 0.14980728674278235,
      "grad_norm": 518.0875244140625,
      "learning_rate": 4.9622944077924106e-05,
      "loss": 138.3091,
      "step": 18540
    },
    {
      "epoch": 0.14988808894706648,
      "grad_norm": 1520.2349853515625,
      "learning_rate": 4.9621722967820184e-05,
      "loss": 134.8868,
      "step": 18550
    },
    {
      "epoch": 0.14996889115135062,
      "grad_norm": 1127.28662109375,
      "learning_rate": 4.962049989867877e-05,
      "loss": 149.5082,
      "step": 18560
    },
    {
      "epoch": 0.15004969335563473,
      "grad_norm": 1920.3206787109375,
      "learning_rate": 4.961927487059716e-05,
      "loss": 206.6959,
      "step": 18570
    },
    {
      "epoch": 0.15013049555991886,
      "grad_norm": 985.6328735351562,
      "learning_rate": 4.961804788367285e-05,
      "loss": 147.696,
      "step": 18580
    },
    {
      "epoch": 0.150211297764203,
      "grad_norm": 1025.6527099609375,
      "learning_rate": 4.961681893800344e-05,
      "loss": 142.4606,
      "step": 18590
    },
    {
      "epoch": 0.15029209996848714,
      "grad_norm": 1423.64794921875,
      "learning_rate": 4.961558803368673e-05,
      "loss": 171.7357,
      "step": 18600
    },
    {
      "epoch": 0.15037290217277127,
      "grad_norm": 1430.884765625,
      "learning_rate": 4.961435517082065e-05,
      "loss": 176.0427,
      "step": 18610
    },
    {
      "epoch": 0.1504537043770554,
      "grad_norm": 664.1461181640625,
      "learning_rate": 4.9613120349503286e-05,
      "loss": 93.5966,
      "step": 18620
    },
    {
      "epoch": 0.15053450658133954,
      "grad_norm": 1300.7425537109375,
      "learning_rate": 4.961188356983291e-05,
      "loss": 171.2061,
      "step": 18630
    },
    {
      "epoch": 0.15061530878562368,
      "grad_norm": 1404.8843994140625,
      "learning_rate": 4.9610644831907896e-05,
      "loss": 156.9324,
      "step": 18640
    },
    {
      "epoch": 0.15069611098990782,
      "grad_norm": 787.6842651367188,
      "learning_rate": 4.960940413582683e-05,
      "loss": 151.5972,
      "step": 18650
    },
    {
      "epoch": 0.15077691319419192,
      "grad_norm": 926.42333984375,
      "learning_rate": 4.960816148168842e-05,
      "loss": 111.1213,
      "step": 18660
    },
    {
      "epoch": 0.15085771539847606,
      "grad_norm": 650.7733154296875,
      "learning_rate": 4.9606916869591527e-05,
      "loss": 143.3633,
      "step": 18670
    },
    {
      "epoch": 0.1509385176027602,
      "grad_norm": 1304.4024658203125,
      "learning_rate": 4.960567029963519e-05,
      "loss": 164.9596,
      "step": 18680
    },
    {
      "epoch": 0.15101931980704433,
      "grad_norm": 1975.5609130859375,
      "learning_rate": 4.9604421771918594e-05,
      "loss": 159.01,
      "step": 18690
    },
    {
      "epoch": 0.15110012201132847,
      "grad_norm": 786.7325439453125,
      "learning_rate": 4.960317128654108e-05,
      "loss": 131.1655,
      "step": 18700
    },
    {
      "epoch": 0.1511809242156126,
      "grad_norm": 1006.70654296875,
      "learning_rate": 4.9601918843602145e-05,
      "loss": 172.4386,
      "step": 18710
    },
    {
      "epoch": 0.15126172641989674,
      "grad_norm": 913.4758911132812,
      "learning_rate": 4.960066444320143e-05,
      "loss": 123.8537,
      "step": 18720
    },
    {
      "epoch": 0.15134252862418088,
      "grad_norm": 1134.485107421875,
      "learning_rate": 4.959940808543875e-05,
      "loss": 143.2191,
      "step": 18730
    },
    {
      "epoch": 0.151423330828465,
      "grad_norm": 1133.67431640625,
      "learning_rate": 4.959814977041406e-05,
      "loss": 111.9317,
      "step": 18740
    },
    {
      "epoch": 0.15150413303274912,
      "grad_norm": 982.4358520507812,
      "learning_rate": 4.9596889498227486e-05,
      "loss": 160.2977,
      "step": 18750
    },
    {
      "epoch": 0.15158493523703326,
      "grad_norm": 816.73681640625,
      "learning_rate": 4.9595627268979294e-05,
      "loss": 128.3574,
      "step": 18760
    },
    {
      "epoch": 0.1516657374413174,
      "grad_norm": 720.4200439453125,
      "learning_rate": 4.9594363082769925e-05,
      "loss": 132.2612,
      "step": 18770
    },
    {
      "epoch": 0.15174653964560153,
      "grad_norm": 1521.4068603515625,
      "learning_rate": 4.959309693969996e-05,
      "loss": 129.5006,
      "step": 18780
    },
    {
      "epoch": 0.15182734184988567,
      "grad_norm": 856.1842041015625,
      "learning_rate": 4.959182883987012e-05,
      "loss": 145.0418,
      "step": 18790
    },
    {
      "epoch": 0.1519081440541698,
      "grad_norm": 1037.7059326171875,
      "learning_rate": 4.959055878338134e-05,
      "loss": 156.9543,
      "step": 18800
    },
    {
      "epoch": 0.15198894625845394,
      "grad_norm": 499.3943786621094,
      "learning_rate": 4.9589286770334654e-05,
      "loss": 127.4943,
      "step": 18810
    },
    {
      "epoch": 0.15206974846273807,
      "grad_norm": 1569.54736328125,
      "learning_rate": 4.9588012800831264e-05,
      "loss": 186.4054,
      "step": 18820
    },
    {
      "epoch": 0.15215055066702218,
      "grad_norm": 1288.7344970703125,
      "learning_rate": 4.9586736874972535e-05,
      "loss": 134.4534,
      "step": 18830
    },
    {
      "epoch": 0.15223135287130632,
      "grad_norm": 981.8977661132812,
      "learning_rate": 4.958545899285999e-05,
      "loss": 147.6222,
      "step": 18840
    },
    {
      "epoch": 0.15231215507559046,
      "grad_norm": 1461.2166748046875,
      "learning_rate": 4.958417915459531e-05,
      "loss": 136.6087,
      "step": 18850
    },
    {
      "epoch": 0.1523929572798746,
      "grad_norm": 1042.28759765625,
      "learning_rate": 4.958289736028032e-05,
      "loss": 125.8117,
      "step": 18860
    },
    {
      "epoch": 0.15247375948415873,
      "grad_norm": 1394.02099609375,
      "learning_rate": 4.958161361001701e-05,
      "loss": 106.0165,
      "step": 18870
    },
    {
      "epoch": 0.15255456168844286,
      "grad_norm": 1479.97509765625,
      "learning_rate": 4.9580327903907514e-05,
      "loss": 135.8925,
      "step": 18880
    },
    {
      "epoch": 0.152635363892727,
      "grad_norm": 983.6040649414062,
      "learning_rate": 4.957904024205414e-05,
      "loss": 182.2857,
      "step": 18890
    },
    {
      "epoch": 0.15271616609701114,
      "grad_norm": 952.5586547851562,
      "learning_rate": 4.957775062455933e-05,
      "loss": 114.4358,
      "step": 18900
    },
    {
      "epoch": 0.15279696830129527,
      "grad_norm": 1429.8126220703125,
      "learning_rate": 4.95764590515257e-05,
      "loss": 155.2205,
      "step": 18910
    },
    {
      "epoch": 0.15287777050557938,
      "grad_norm": 1141.6102294921875,
      "learning_rate": 4.957516552305602e-05,
      "loss": 196.4725,
      "step": 18920
    },
    {
      "epoch": 0.15295857270986352,
      "grad_norm": 971.5953369140625,
      "learning_rate": 4.957387003925321e-05,
      "loss": 143.6367,
      "step": 18930
    },
    {
      "epoch": 0.15303937491414765,
      "grad_norm": 986.298095703125,
      "learning_rate": 4.9572572600220323e-05,
      "loss": 161.2457,
      "step": 18940
    },
    {
      "epoch": 0.1531201771184318,
      "grad_norm": 618.5545654296875,
      "learning_rate": 4.957127320606062e-05,
      "loss": 98.7228,
      "step": 18950
    },
    {
      "epoch": 0.15320097932271592,
      "grad_norm": 1332.11572265625,
      "learning_rate": 4.956997185687747e-05,
      "loss": 119.0759,
      "step": 18960
    },
    {
      "epoch": 0.15328178152700006,
      "grad_norm": 1110.71923828125,
      "learning_rate": 4.9568668552774424e-05,
      "loss": 157.8614,
      "step": 18970
    },
    {
      "epoch": 0.1533625837312842,
      "grad_norm": 567.2537841796875,
      "learning_rate": 4.956736329385517e-05,
      "loss": 127.0969,
      "step": 18980
    },
    {
      "epoch": 0.15344338593556833,
      "grad_norm": 828.8348999023438,
      "learning_rate": 4.9566056080223574e-05,
      "loss": 163.5427,
      "step": 18990
    },
    {
      "epoch": 0.15352418813985247,
      "grad_norm": 1140.49853515625,
      "learning_rate": 4.956474691198363e-05,
      "loss": 132.9715,
      "step": 19000
    },
    {
      "epoch": 0.15360499034413658,
      "grad_norm": 958.723388671875,
      "learning_rate": 4.956343578923952e-05,
      "loss": 136.1955,
      "step": 19010
    },
    {
      "epoch": 0.1536857925484207,
      "grad_norm": 990.290771484375,
      "learning_rate": 4.956212271209555e-05,
      "loss": 133.8689,
      "step": 19020
    },
    {
      "epoch": 0.15376659475270485,
      "grad_norm": 1795.55908203125,
      "learning_rate": 4.956080768065621e-05,
      "loss": 141.3235,
      "step": 19030
    },
    {
      "epoch": 0.15384739695698899,
      "grad_norm": 794.1259765625,
      "learning_rate": 4.9559490695026113e-05,
      "loss": 124.8279,
      "step": 19040
    },
    {
      "epoch": 0.15392819916127312,
      "grad_norm": 1143.3475341796875,
      "learning_rate": 4.955817175531005e-05,
      "loss": 157.4777,
      "step": 19050
    },
    {
      "epoch": 0.15400900136555726,
      "grad_norm": 725.7691650390625,
      "learning_rate": 4.9556850861612976e-05,
      "loss": 152.8616,
      "step": 19060
    },
    {
      "epoch": 0.1540898035698414,
      "grad_norm": 829.7158813476562,
      "learning_rate": 4.955552801403998e-05,
      "loss": 127.6073,
      "step": 19070
    },
    {
      "epoch": 0.15417060577412553,
      "grad_norm": 836.8492431640625,
      "learning_rate": 4.9554203212696304e-05,
      "loss": 128.8572,
      "step": 19080
    },
    {
      "epoch": 0.15425140797840964,
      "grad_norm": 939.9307250976562,
      "learning_rate": 4.9552876457687374e-05,
      "loss": 132.6982,
      "step": 19090
    },
    {
      "epoch": 0.15433221018269377,
      "grad_norm": 1570.5841064453125,
      "learning_rate": 4.955154774911875e-05,
      "loss": 101.3377,
      "step": 19100
    },
    {
      "epoch": 0.1544130123869779,
      "grad_norm": 608.4269409179688,
      "learning_rate": 4.955021708709614e-05,
      "loss": 138.4897,
      "step": 19110
    },
    {
      "epoch": 0.15449381459126205,
      "grad_norm": 1717.3123779296875,
      "learning_rate": 4.9548884471725434e-05,
      "loss": 183.6861,
      "step": 19120
    },
    {
      "epoch": 0.15457461679554618,
      "grad_norm": 721.2388305664062,
      "learning_rate": 4.9547549903112654e-05,
      "loss": 183.5123,
      "step": 19130
    },
    {
      "epoch": 0.15465541899983032,
      "grad_norm": 1664.793701171875,
      "learning_rate": 4.954621338136398e-05,
      "loss": 160.0894,
      "step": 19140
    },
    {
      "epoch": 0.15473622120411445,
      "grad_norm": 810.6116333007812,
      "learning_rate": 4.954487490658577e-05,
      "loss": 150.3457,
      "step": 19150
    },
    {
      "epoch": 0.1548170234083986,
      "grad_norm": 1055.0728759765625,
      "learning_rate": 4.95435344788845e-05,
      "loss": 168.4552,
      "step": 19160
    },
    {
      "epoch": 0.15489782561268273,
      "grad_norm": 878.2693481445312,
      "learning_rate": 4.954219209836684e-05,
      "loss": 123.0568,
      "step": 19170
    },
    {
      "epoch": 0.15497862781696684,
      "grad_norm": 971.3629760742188,
      "learning_rate": 4.954084776513957e-05,
      "loss": 112.7831,
      "step": 19180
    },
    {
      "epoch": 0.15505943002125097,
      "grad_norm": 1250.1171875,
      "learning_rate": 4.953950147930969e-05,
      "loss": 128.6498,
      "step": 19190
    },
    {
      "epoch": 0.1551402322255351,
      "grad_norm": 1516.2861328125,
      "learning_rate": 4.9538153240984286e-05,
      "loss": 135.834,
      "step": 19200
    },
    {
      "epoch": 0.15522103442981924,
      "grad_norm": 917.4509887695312,
      "learning_rate": 4.953680305027065e-05,
      "loss": 199.9036,
      "step": 19210
    },
    {
      "epoch": 0.15530183663410338,
      "grad_norm": 1195.2305908203125,
      "learning_rate": 4.9535450907276204e-05,
      "loss": 167.8621,
      "step": 19220
    },
    {
      "epoch": 0.15538263883838752,
      "grad_norm": 1185.2828369140625,
      "learning_rate": 4.953409681210853e-05,
      "loss": 150.506,
      "step": 19230
    },
    {
      "epoch": 0.15546344104267165,
      "grad_norm": 1028.5968017578125,
      "learning_rate": 4.9532740764875377e-05,
      "loss": 160.7915,
      "step": 19240
    },
    {
      "epoch": 0.1555442432469558,
      "grad_norm": 1641.87451171875,
      "learning_rate": 4.953138276568462e-05,
      "loss": 167.9964,
      "step": 19250
    },
    {
      "epoch": 0.1556250454512399,
      "grad_norm": 676.5661010742188,
      "learning_rate": 4.953002281464432e-05,
      "loss": 168.5508,
      "step": 19260
    },
    {
      "epoch": 0.15570584765552403,
      "grad_norm": 581.885009765625,
      "learning_rate": 4.952866091186269e-05,
      "loss": 138.8725,
      "step": 19270
    },
    {
      "epoch": 0.15578664985980817,
      "grad_norm": 1236.295166015625,
      "learning_rate": 4.952729705744808e-05,
      "loss": 123.4911,
      "step": 19280
    },
    {
      "epoch": 0.1558674520640923,
      "grad_norm": 888.4049682617188,
      "learning_rate": 4.9525931251509e-05,
      "loss": 138.8721,
      "step": 19290
    },
    {
      "epoch": 0.15594825426837644,
      "grad_norm": 656.3297119140625,
      "learning_rate": 4.9524563494154145e-05,
      "loss": 161.5853,
      "step": 19300
    },
    {
      "epoch": 0.15602905647266058,
      "grad_norm": 899.3927612304688,
      "learning_rate": 4.952319378549232e-05,
      "loss": 130.1569,
      "step": 19310
    },
    {
      "epoch": 0.1561098586769447,
      "grad_norm": 1051.7786865234375,
      "learning_rate": 4.95218221256325e-05,
      "loss": 155.7519,
      "step": 19320
    },
    {
      "epoch": 0.15619066088122885,
      "grad_norm": 351.5675048828125,
      "learning_rate": 4.952044851468385e-05,
      "loss": 118.1119,
      "step": 19330
    },
    {
      "epoch": 0.15627146308551298,
      "grad_norm": 648.6687622070312,
      "learning_rate": 4.951907295275563e-05,
      "loss": 186.3808,
      "step": 19340
    },
    {
      "epoch": 0.1563522652897971,
      "grad_norm": 737.3397827148438,
      "learning_rate": 4.951769543995731e-05,
      "loss": 164.0862,
      "step": 19350
    },
    {
      "epoch": 0.15643306749408123,
      "grad_norm": 1181.345458984375,
      "learning_rate": 4.951631597639849e-05,
      "loss": 95.8942,
      "step": 19360
    },
    {
      "epoch": 0.15651386969836537,
      "grad_norm": 1376.4288330078125,
      "learning_rate": 4.9514934562188915e-05,
      "loss": 155.4059,
      "step": 19370
    },
    {
      "epoch": 0.1565946719026495,
      "grad_norm": 984.8242797851562,
      "learning_rate": 4.951355119743851e-05,
      "loss": 136.2228,
      "step": 19380
    },
    {
      "epoch": 0.15667547410693364,
      "grad_norm": 1276.46142578125,
      "learning_rate": 4.9512165882257335e-05,
      "loss": 134.6281,
      "step": 19390
    },
    {
      "epoch": 0.15675627631121777,
      "grad_norm": 855.7882690429688,
      "learning_rate": 4.9510778616755616e-05,
      "loss": 127.1215,
      "step": 19400
    },
    {
      "epoch": 0.1568370785155019,
      "grad_norm": 1180.2899169921875,
      "learning_rate": 4.9509389401043735e-05,
      "loss": 157.9484,
      "step": 19410
    },
    {
      "epoch": 0.15691788071978605,
      "grad_norm": 920.9578857421875,
      "learning_rate": 4.950799823523222e-05,
      "loss": 140.1964,
      "step": 19420
    },
    {
      "epoch": 0.15699868292407018,
      "grad_norm": 1415.94091796875,
      "learning_rate": 4.950660511943176e-05,
      "loss": 135.1266,
      "step": 19430
    },
    {
      "epoch": 0.1570794851283543,
      "grad_norm": 396.073974609375,
      "learning_rate": 4.95052100537532e-05,
      "loss": 144.9773,
      "step": 19440
    },
    {
      "epoch": 0.15716028733263843,
      "grad_norm": 716.8720092773438,
      "learning_rate": 4.950381303830755e-05,
      "loss": 118.7841,
      "step": 19450
    },
    {
      "epoch": 0.15724108953692256,
      "grad_norm": 798.32666015625,
      "learning_rate": 4.950241407320594e-05,
      "loss": 136.819,
      "step": 19460
    },
    {
      "epoch": 0.1573218917412067,
      "grad_norm": 1627.865966796875,
      "learning_rate": 4.95010131585597e-05,
      "loss": 155.6983,
      "step": 19470
    },
    {
      "epoch": 0.15740269394549083,
      "grad_norm": 2705.7021484375,
      "learning_rate": 4.9499610294480284e-05,
      "loss": 151.8552,
      "step": 19480
    },
    {
      "epoch": 0.15748349614977497,
      "grad_norm": 970.4476928710938,
      "learning_rate": 4.9498205481079315e-05,
      "loss": 154.0217,
      "step": 19490
    },
    {
      "epoch": 0.1575642983540591,
      "grad_norm": 775.9421997070312,
      "learning_rate": 4.949679871846857e-05,
      "loss": 171.9853,
      "step": 19500
    },
    {
      "epoch": 0.15764510055834324,
      "grad_norm": 736.22314453125,
      "learning_rate": 4.949539000675998e-05,
      "loss": 115.1917,
      "step": 19510
    },
    {
      "epoch": 0.15772590276262735,
      "grad_norm": 773.5701293945312,
      "learning_rate": 4.9493979346065624e-05,
      "loss": 143.0012,
      "step": 19520
    },
    {
      "epoch": 0.1578067049669115,
      "grad_norm": 1225.4534912109375,
      "learning_rate": 4.9492566736497744e-05,
      "loss": 146.8361,
      "step": 19530
    },
    {
      "epoch": 0.15788750717119562,
      "grad_norm": 849.607421875,
      "learning_rate": 4.949115217816873e-05,
      "loss": 154.6573,
      "step": 19540
    },
    {
      "epoch": 0.15796830937547976,
      "grad_norm": 1171.7467041015625,
      "learning_rate": 4.948973567119114e-05,
      "loss": 143.7283,
      "step": 19550
    },
    {
      "epoch": 0.1580491115797639,
      "grad_norm": 1184.739013671875,
      "learning_rate": 4.9488317215677673e-05,
      "loss": 147.067,
      "step": 19560
    },
    {
      "epoch": 0.15812991378404803,
      "grad_norm": 671.5252075195312,
      "learning_rate": 4.948689681174119e-05,
      "loss": 134.3241,
      "step": 19570
    },
    {
      "epoch": 0.15821071598833217,
      "grad_norm": 1588.4747314453125,
      "learning_rate": 4.948547445949471e-05,
      "loss": 132.0108,
      "step": 19580
    },
    {
      "epoch": 0.1582915181926163,
      "grad_norm": 1395.05712890625,
      "learning_rate": 4.94840501590514e-05,
      "loss": 132.0019,
      "step": 19590
    },
    {
      "epoch": 0.15837232039690044,
      "grad_norm": 1854.0704345703125,
      "learning_rate": 4.948262391052458e-05,
      "loss": 113.3447,
      "step": 19600
    },
    {
      "epoch": 0.15845312260118455,
      "grad_norm": 1134.071044921875,
      "learning_rate": 4.948119571402775e-05,
      "loss": 133.9102,
      "step": 19610
    },
    {
      "epoch": 0.15853392480546868,
      "grad_norm": 1435.5302734375,
      "learning_rate": 4.947976556967452e-05,
      "loss": 142.5358,
      "step": 19620
    },
    {
      "epoch": 0.15861472700975282,
      "grad_norm": 672.7803344726562,
      "learning_rate": 4.947833347757869e-05,
      "loss": 104.7007,
      "step": 19630
    },
    {
      "epoch": 0.15869552921403696,
      "grad_norm": 597.4839477539062,
      "learning_rate": 4.9476899437854205e-05,
      "loss": 134.072,
      "step": 19640
    },
    {
      "epoch": 0.1587763314183211,
      "grad_norm": 1603.63525390625,
      "learning_rate": 4.9475463450615175e-05,
      "loss": 115.9036,
      "step": 19650
    },
    {
      "epoch": 0.15885713362260523,
      "grad_norm": 663.64306640625,
      "learning_rate": 4.9474025515975835e-05,
      "loss": 112.9661,
      "step": 19660
    },
    {
      "epoch": 0.15893793582688936,
      "grad_norm": 1277.085205078125,
      "learning_rate": 4.947258563405061e-05,
      "loss": 138.1168,
      "step": 19670
    },
    {
      "epoch": 0.1590187380311735,
      "grad_norm": 1482.751708984375,
      "learning_rate": 4.947114380495406e-05,
      "loss": 194.949,
      "step": 19680
    },
    {
      "epoch": 0.15909954023545764,
      "grad_norm": 1446.81787109375,
      "learning_rate": 4.94697000288009e-05,
      "loss": 145.5572,
      "step": 19690
    },
    {
      "epoch": 0.15918034243974175,
      "grad_norm": 932.5813598632812,
      "learning_rate": 4.946825430570602e-05,
      "loss": 145.6906,
      "step": 19700
    },
    {
      "epoch": 0.15926114464402588,
      "grad_norm": 1143.1041259765625,
      "learning_rate": 4.946680663578443e-05,
      "loss": 151.2845,
      "step": 19710
    },
    {
      "epoch": 0.15934194684831002,
      "grad_norm": 1063.902587890625,
      "learning_rate": 4.9465357019151325e-05,
      "loss": 155.4715,
      "step": 19720
    },
    {
      "epoch": 0.15942274905259415,
      "grad_norm": 945.45556640625,
      "learning_rate": 4.946390545592204e-05,
      "loss": 188.0311,
      "step": 19730
    },
    {
      "epoch": 0.1595035512568783,
      "grad_norm": 1733.0693359375,
      "learning_rate": 4.9462451946212085e-05,
      "loss": 150.3059,
      "step": 19740
    },
    {
      "epoch": 0.15958435346116243,
      "grad_norm": 934.3817749023438,
      "learning_rate": 4.946099649013708e-05,
      "loss": 115.0981,
      "step": 19750
    },
    {
      "epoch": 0.15966515566544656,
      "grad_norm": 756.0308227539062,
      "learning_rate": 4.945953908781286e-05,
      "loss": 126.3076,
      "step": 19760
    },
    {
      "epoch": 0.1597459578697307,
      "grad_norm": 646.1735229492188,
      "learning_rate": 4.945807973935536e-05,
      "loss": 141.5272,
      "step": 19770
    },
    {
      "epoch": 0.1598267600740148,
      "grad_norm": 900.1630249023438,
      "learning_rate": 4.94566184448807e-05,
      "loss": 169.0281,
      "step": 19780
    },
    {
      "epoch": 0.15990756227829894,
      "grad_norm": 1312.0203857421875,
      "learning_rate": 4.945515520450515e-05,
      "loss": 168.3111,
      "step": 19790
    },
    {
      "epoch": 0.15998836448258308,
      "grad_norm": 824.4857177734375,
      "learning_rate": 4.9453690018345144e-05,
      "loss": 160.6633,
      "step": 19800
    },
    {
      "epoch": 0.16006916668686721,
      "grad_norm": 606.3312377929688,
      "learning_rate": 4.945222288651724e-05,
      "loss": 167.2263,
      "step": 19810
    },
    {
      "epoch": 0.16014996889115135,
      "grad_norm": 1857.02685546875,
      "learning_rate": 4.945075380913819e-05,
      "loss": 190.6399,
      "step": 19820
    },
    {
      "epoch": 0.1602307710954355,
      "grad_norm": 1485.077392578125,
      "learning_rate": 4.944928278632487e-05,
      "loss": 156.7114,
      "step": 19830
    },
    {
      "epoch": 0.16031157329971962,
      "grad_norm": 824.557373046875,
      "learning_rate": 4.944780981819433e-05,
      "loss": 143.3107,
      "step": 19840
    },
    {
      "epoch": 0.16039237550400376,
      "grad_norm": 987.8336791992188,
      "learning_rate": 4.944633490486376e-05,
      "loss": 100.8076,
      "step": 19850
    },
    {
      "epoch": 0.1604731777082879,
      "grad_norm": 1033.8603515625,
      "learning_rate": 4.944485804645052e-05,
      "loss": 162.6909,
      "step": 19860
    },
    {
      "epoch": 0.160553979912572,
      "grad_norm": 1208.977783203125,
      "learning_rate": 4.9443379243072094e-05,
      "loss": 132.4686,
      "step": 19870
    },
    {
      "epoch": 0.16063478211685614,
      "grad_norm": 719.5220947265625,
      "learning_rate": 4.944189849484618e-05,
      "loss": 141.4025,
      "step": 19880
    },
    {
      "epoch": 0.16071558432114028,
      "grad_norm": 1385.524658203125,
      "learning_rate": 4.9440415801890566e-05,
      "loss": 182.8163,
      "step": 19890
    },
    {
      "epoch": 0.1607963865254244,
      "grad_norm": 1521.5345458984375,
      "learning_rate": 4.9438931164323236e-05,
      "loss": 110.8893,
      "step": 19900
    },
    {
      "epoch": 0.16087718872970855,
      "grad_norm": 931.8245849609375,
      "learning_rate": 4.9437444582262316e-05,
      "loss": 141.6139,
      "step": 19910
    },
    {
      "epoch": 0.16095799093399268,
      "grad_norm": 1338.6199951171875,
      "learning_rate": 4.9435956055826083e-05,
      "loss": 167.0462,
      "step": 19920
    },
    {
      "epoch": 0.16103879313827682,
      "grad_norm": 592.0842895507812,
      "learning_rate": 4.943446558513297e-05,
      "loss": 104.6128,
      "step": 19930
    },
    {
      "epoch": 0.16111959534256096,
      "grad_norm": 765.056640625,
      "learning_rate": 4.943297317030156e-05,
      "loss": 136.0432,
      "step": 19940
    },
    {
      "epoch": 0.16120039754684506,
      "grad_norm": 896.3486328125,
      "learning_rate": 4.943147881145063e-05,
      "loss": 152.3156,
      "step": 19950
    },
    {
      "epoch": 0.1612811997511292,
      "grad_norm": 669.0186157226562,
      "learning_rate": 4.942998250869904e-05,
      "loss": 146.4306,
      "step": 19960
    },
    {
      "epoch": 0.16136200195541334,
      "grad_norm": 738.7496948242188,
      "learning_rate": 4.9428484262165865e-05,
      "loss": 167.2415,
      "step": 19970
    },
    {
      "epoch": 0.16144280415969747,
      "grad_norm": 1352.85009765625,
      "learning_rate": 4.9426984071970305e-05,
      "loss": 168.2593,
      "step": 19980
    },
    {
      "epoch": 0.1615236063639816,
      "grad_norm": 1093.584228515625,
      "learning_rate": 4.942548193823173e-05,
      "loss": 186.0099,
      "step": 19990
    },
    {
      "epoch": 0.16160440856826574,
      "grad_norm": 1222.2086181640625,
      "learning_rate": 4.942397786106965e-05,
      "loss": 159.3606,
      "step": 20000
    },
    {
      "epoch": 0.16168521077254988,
      "grad_norm": 1016.3326416015625,
      "learning_rate": 4.942247184060375e-05,
      "loss": 121.1088,
      "step": 20010
    },
    {
      "epoch": 0.16176601297683402,
      "grad_norm": 760.5179443359375,
      "learning_rate": 4.942096387695385e-05,
      "loss": 126.0567,
      "step": 20020
    },
    {
      "epoch": 0.16184681518111815,
      "grad_norm": 2420.37841796875,
      "learning_rate": 4.941945397023993e-05,
      "loss": 142.4873,
      "step": 20030
    },
    {
      "epoch": 0.16192761738540226,
      "grad_norm": 1585.40234375,
      "learning_rate": 4.9417942120582114e-05,
      "loss": 128.2617,
      "step": 20040
    },
    {
      "epoch": 0.1620084195896864,
      "grad_norm": 1405.9205322265625,
      "learning_rate": 4.941642832810072e-05,
      "loss": 123.314,
      "step": 20050
    },
    {
      "epoch": 0.16208922179397053,
      "grad_norm": 827.2435913085938,
      "learning_rate": 4.9414912592916185e-05,
      "loss": 143.6495,
      "step": 20060
    },
    {
      "epoch": 0.16217002399825467,
      "grad_norm": 742.4790649414062,
      "learning_rate": 4.9413394915149094e-05,
      "loss": 200.291,
      "step": 20070
    },
    {
      "epoch": 0.1622508262025388,
      "grad_norm": 1240.279296875,
      "learning_rate": 4.9411875294920215e-05,
      "loss": 165.2704,
      "step": 20080
    },
    {
      "epoch": 0.16233162840682294,
      "grad_norm": 671.4786987304688,
      "learning_rate": 4.9410353732350455e-05,
      "loss": 158.4867,
      "step": 20090
    },
    {
      "epoch": 0.16241243061110708,
      "grad_norm": 609.6265869140625,
      "learning_rate": 4.940883022756088e-05,
      "loss": 164.4025,
      "step": 20100
    },
    {
      "epoch": 0.16249323281539121,
      "grad_norm": 702.2029418945312,
      "learning_rate": 4.94073047806727e-05,
      "loss": 143.9458,
      "step": 20110
    },
    {
      "epoch": 0.16257403501967535,
      "grad_norm": 658.32763671875,
      "learning_rate": 4.94057773918073e-05,
      "loss": 116.2497,
      "step": 20120
    },
    {
      "epoch": 0.16265483722395946,
      "grad_norm": 618.864990234375,
      "learning_rate": 4.940424806108619e-05,
      "loss": 113.3011,
      "step": 20130
    },
    {
      "epoch": 0.1627356394282436,
      "grad_norm": 646.5340576171875,
      "learning_rate": 4.9402716788631073e-05,
      "loss": 209.5348,
      "step": 20140
    },
    {
      "epoch": 0.16281644163252773,
      "grad_norm": 794.1364135742188,
      "learning_rate": 4.940118357456377e-05,
      "loss": 112.0908,
      "step": 20150
    },
    {
      "epoch": 0.16289724383681187,
      "grad_norm": 1476.8046875,
      "learning_rate": 4.939964841900627e-05,
      "loss": 156.0068,
      "step": 20160
    },
    {
      "epoch": 0.162978046041096,
      "grad_norm": 1192.75146484375,
      "learning_rate": 4.939811132208073e-05,
      "loss": 208.6819,
      "step": 20170
    },
    {
      "epoch": 0.16305884824538014,
      "grad_norm": 1222.39208984375,
      "learning_rate": 4.939657228390945e-05,
      "loss": 127.3492,
      "step": 20180
    },
    {
      "epoch": 0.16313965044966428,
      "grad_norm": 1381.251708984375,
      "learning_rate": 4.939503130461487e-05,
      "loss": 141.149,
      "step": 20190
    },
    {
      "epoch": 0.1632204526539484,
      "grad_norm": 948.350341796875,
      "learning_rate": 4.9393488384319605e-05,
      "loss": 146.2455,
      "step": 20200
    },
    {
      "epoch": 0.16330125485823252,
      "grad_norm": 1100.5667724609375,
      "learning_rate": 4.939194352314643e-05,
      "loss": 134.0395,
      "step": 20210
    },
    {
      "epoch": 0.16338205706251666,
      "grad_norm": 1312.8271484375,
      "learning_rate": 4.939039672121825e-05,
      "loss": 162.6504,
      "step": 20220
    },
    {
      "epoch": 0.1634628592668008,
      "grad_norm": 742.9317626953125,
      "learning_rate": 4.938884797865814e-05,
      "loss": 119.811,
      "step": 20230
    },
    {
      "epoch": 0.16354366147108493,
      "grad_norm": 1479.2552490234375,
      "learning_rate": 4.938729729558932e-05,
      "loss": 127.0003,
      "step": 20240
    },
    {
      "epoch": 0.16362446367536906,
      "grad_norm": 2271.06982421875,
      "learning_rate": 4.938574467213518e-05,
      "loss": 180.906,
      "step": 20250
    },
    {
      "epoch": 0.1637052658796532,
      "grad_norm": 848.5759887695312,
      "learning_rate": 4.938419010841925e-05,
      "loss": 108.292,
      "step": 20260
    },
    {
      "epoch": 0.16378606808393734,
      "grad_norm": 865.2387084960938,
      "learning_rate": 4.938263360456523e-05,
      "loss": 146.3953,
      "step": 20270
    },
    {
      "epoch": 0.16386687028822147,
      "grad_norm": 1448.966552734375,
      "learning_rate": 4.938107516069694e-05,
      "loss": 197.9948,
      "step": 20280
    },
    {
      "epoch": 0.1639476724925056,
      "grad_norm": 856.9296875,
      "learning_rate": 4.9379514776938405e-05,
      "loss": 120.4137,
      "step": 20290
    },
    {
      "epoch": 0.16402847469678972,
      "grad_norm": 1423.47607421875,
      "learning_rate": 4.9377952453413765e-05,
      "loss": 147.8874,
      "step": 20300
    },
    {
      "epoch": 0.16410927690107385,
      "grad_norm": 1550.7271728515625,
      "learning_rate": 4.9376388190247324e-05,
      "loss": 153.0022,
      "step": 20310
    },
    {
      "epoch": 0.164190079105358,
      "grad_norm": 1133.5203857421875,
      "learning_rate": 4.937482198756355e-05,
      "loss": 169.8442,
      "step": 20320
    },
    {
      "epoch": 0.16427088130964212,
      "grad_norm": 921.5442504882812,
      "learning_rate": 4.937325384548705e-05,
      "loss": 179.4838,
      "step": 20330
    },
    {
      "epoch": 0.16435168351392626,
      "grad_norm": 920.5408935546875,
      "learning_rate": 4.9371683764142615e-05,
      "loss": 108.6345,
      "step": 20340
    },
    {
      "epoch": 0.1644324857182104,
      "grad_norm": 1286.8828125,
      "learning_rate": 4.9370111743655145e-05,
      "loss": 148.3994,
      "step": 20350
    },
    {
      "epoch": 0.16451328792249453,
      "grad_norm": 918.86181640625,
      "learning_rate": 4.9368537784149724e-05,
      "loss": 139.1599,
      "step": 20360
    },
    {
      "epoch": 0.16459409012677867,
      "grad_norm": 1502.0557861328125,
      "learning_rate": 4.936696188575159e-05,
      "loss": 164.7047,
      "step": 20370
    },
    {
      "epoch": 0.16467489233106278,
      "grad_norm": 921.6201171875,
      "learning_rate": 4.9365384048586125e-05,
      "loss": 146.2348,
      "step": 20380
    },
    {
      "epoch": 0.1647556945353469,
      "grad_norm": 761.9798583984375,
      "learning_rate": 4.936380427277888e-05,
      "loss": 122.8679,
      "step": 20390
    },
    {
      "epoch": 0.16483649673963105,
      "grad_norm": 1140.500732421875,
      "learning_rate": 4.936222255845554e-05,
      "loss": 120.364,
      "step": 20400
    },
    {
      "epoch": 0.16491729894391519,
      "grad_norm": 1554.974609375,
      "learning_rate": 4.936063890574196e-05,
      "loss": 170.4009,
      "step": 20410
    },
    {
      "epoch": 0.16499810114819932,
      "grad_norm": 756.5431518554688,
      "learning_rate": 4.935905331476414e-05,
      "loss": 124.2604,
      "step": 20420
    },
    {
      "epoch": 0.16507890335248346,
      "grad_norm": 759.2258911132812,
      "learning_rate": 4.9357465785648247e-05,
      "loss": 122.2716,
      "step": 20430
    },
    {
      "epoch": 0.1651597055567676,
      "grad_norm": 1130.5928955078125,
      "learning_rate": 4.935587631852058e-05,
      "loss": 120.7589,
      "step": 20440
    },
    {
      "epoch": 0.16524050776105173,
      "grad_norm": 994.0823364257812,
      "learning_rate": 4.935428491350761e-05,
      "loss": 147.596,
      "step": 20450
    },
    {
      "epoch": 0.16532130996533587,
      "grad_norm": 1129.5496826171875,
      "learning_rate": 4.9352691570735965e-05,
      "loss": 175.5725,
      "step": 20460
    },
    {
      "epoch": 0.16540211216961997,
      "grad_norm": 974.0113525390625,
      "learning_rate": 4.935109629033242e-05,
      "loss": 198.5151,
      "step": 20470
    },
    {
      "epoch": 0.1654829143739041,
      "grad_norm": 1393.72607421875,
      "learning_rate": 4.93494990724239e-05,
      "loss": 142.9287,
      "step": 20480
    },
    {
      "epoch": 0.16556371657818825,
      "grad_norm": 1063.649169921875,
      "learning_rate": 4.934789991713748e-05,
      "loss": 123.0873,
      "step": 20490
    },
    {
      "epoch": 0.16564451878247238,
      "grad_norm": 1238.6739501953125,
      "learning_rate": 4.9346298824600405e-05,
      "loss": 120.6401,
      "step": 20500
    },
    {
      "epoch": 0.16572532098675652,
      "grad_norm": 843.2783203125,
      "learning_rate": 4.934469579494008e-05,
      "loss": 121.2858,
      "step": 20510
    },
    {
      "epoch": 0.16580612319104066,
      "grad_norm": 983.3422241210938,
      "learning_rate": 4.934309082828402e-05,
      "loss": 143.3731,
      "step": 20520
    },
    {
      "epoch": 0.1658869253953248,
      "grad_norm": 959.219970703125,
      "learning_rate": 4.934148392475996e-05,
      "loss": 145.4661,
      "step": 20530
    },
    {
      "epoch": 0.16596772759960893,
      "grad_norm": 733.232666015625,
      "learning_rate": 4.933987508449572e-05,
      "loss": 126.4923,
      "step": 20540
    },
    {
      "epoch": 0.16604852980389306,
      "grad_norm": 853.0502319335938,
      "learning_rate": 4.933826430761933e-05,
      "loss": 153.8978,
      "step": 20550
    },
    {
      "epoch": 0.16612933200817717,
      "grad_norm": 1198.431640625,
      "learning_rate": 4.933665159425895e-05,
      "loss": 166.0935,
      "step": 20560
    },
    {
      "epoch": 0.1662101342124613,
      "grad_norm": 514.8040771484375,
      "learning_rate": 4.933503694454289e-05,
      "loss": 85.8809,
      "step": 20570
    },
    {
      "epoch": 0.16629093641674544,
      "grad_norm": 791.435302734375,
      "learning_rate": 4.9333420358599624e-05,
      "loss": 141.1844,
      "step": 20580
    },
    {
      "epoch": 0.16637173862102958,
      "grad_norm": 1096.3251953125,
      "learning_rate": 4.9331801836557776e-05,
      "loss": 141.8443,
      "step": 20590
    },
    {
      "epoch": 0.16645254082531372,
      "grad_norm": 4899.19921875,
      "learning_rate": 4.9330181378546124e-05,
      "loss": 176.5427,
      "step": 20600
    },
    {
      "epoch": 0.16653334302959785,
      "grad_norm": 453.80072021484375,
      "learning_rate": 4.93285589846936e-05,
      "loss": 143.0679,
      "step": 20610
    },
    {
      "epoch": 0.166614145233882,
      "grad_norm": 1067.961669921875,
      "learning_rate": 4.9326934655129295e-05,
      "loss": 125.034,
      "step": 20620
    },
    {
      "epoch": 0.16669494743816612,
      "grad_norm": 674.2916259765625,
      "learning_rate": 4.932530838998244e-05,
      "loss": 121.1949,
      "step": 20630
    },
    {
      "epoch": 0.16677574964245023,
      "grad_norm": 2042.013671875,
      "learning_rate": 4.9323680189382434e-05,
      "loss": 248.2044,
      "step": 20640
    },
    {
      "epoch": 0.16685655184673437,
      "grad_norm": 972.5521240234375,
      "learning_rate": 4.932205005345882e-05,
      "loss": 152.445,
      "step": 20650
    },
    {
      "epoch": 0.1669373540510185,
      "grad_norm": 659.5952758789062,
      "learning_rate": 4.9320417982341313e-05,
      "loss": 123.7578,
      "step": 20660
    },
    {
      "epoch": 0.16701815625530264,
      "grad_norm": 1101.409912109375,
      "learning_rate": 4.9318783976159765e-05,
      "loss": 157.3332,
      "step": 20670
    },
    {
      "epoch": 0.16709895845958678,
      "grad_norm": 1593.623291015625,
      "learning_rate": 4.931714803504418e-05,
      "loss": 161.4305,
      "step": 20680
    },
    {
      "epoch": 0.1671797606638709,
      "grad_norm": 785.7627563476562,
      "learning_rate": 4.9315510159124734e-05,
      "loss": 133.448,
      "step": 20690
    },
    {
      "epoch": 0.16726056286815505,
      "grad_norm": 1091.3284912109375,
      "learning_rate": 4.931387034853173e-05,
      "loss": 167.9184,
      "step": 20700
    },
    {
      "epoch": 0.16734136507243919,
      "grad_norm": 1226.1614990234375,
      "learning_rate": 4.931222860339565e-05,
      "loss": 127.1648,
      "step": 20710
    },
    {
      "epoch": 0.16742216727672332,
      "grad_norm": 1060.7408447265625,
      "learning_rate": 4.931058492384712e-05,
      "loss": 131.7696,
      "step": 20720
    },
    {
      "epoch": 0.16750296948100743,
      "grad_norm": 821.4359130859375,
      "learning_rate": 4.9308939310016916e-05,
      "loss": 135.4715,
      "step": 20730
    },
    {
      "epoch": 0.16758377168529157,
      "grad_norm": 2822.77197265625,
      "learning_rate": 4.930729176203598e-05,
      "loss": 169.724,
      "step": 20740
    },
    {
      "epoch": 0.1676645738895757,
      "grad_norm": 814.9628295898438,
      "learning_rate": 4.930564228003538e-05,
      "loss": 111.3484,
      "step": 20750
    },
    {
      "epoch": 0.16774537609385984,
      "grad_norm": 1205.373046875,
      "learning_rate": 4.930399086414638e-05,
      "loss": 125.1682,
      "step": 20760
    },
    {
      "epoch": 0.16782617829814397,
      "grad_norm": 874.1157836914062,
      "learning_rate": 4.9302337514500374e-05,
      "loss": 145.6209,
      "step": 20770
    },
    {
      "epoch": 0.1679069805024281,
      "grad_norm": 1192.6041259765625,
      "learning_rate": 4.930068223122889e-05,
      "loss": 154.289,
      "step": 20780
    },
    {
      "epoch": 0.16798778270671225,
      "grad_norm": 1004.2535400390625,
      "learning_rate": 4.929902501446366e-05,
      "loss": 183.3328,
      "step": 20790
    },
    {
      "epoch": 0.16806858491099638,
      "grad_norm": 996.1227416992188,
      "learning_rate": 4.929736586433652e-05,
      "loss": 133.2687,
      "step": 20800
    },
    {
      "epoch": 0.16814938711528052,
      "grad_norm": 676.2308349609375,
      "learning_rate": 4.929570478097949e-05,
      "loss": 149.5055,
      "step": 20810
    },
    {
      "epoch": 0.16823018931956463,
      "grad_norm": 1042.8621826171875,
      "learning_rate": 4.9294041764524726e-05,
      "loss": 126.4842,
      "step": 20820
    },
    {
      "epoch": 0.16831099152384876,
      "grad_norm": 1661.6201171875,
      "learning_rate": 4.9292376815104566e-05,
      "loss": 133.8418,
      "step": 20830
    },
    {
      "epoch": 0.1683917937281329,
      "grad_norm": 1082.744140625,
      "learning_rate": 4.929070993285146e-05,
      "loss": 175.319,
      "step": 20840
    },
    {
      "epoch": 0.16847259593241704,
      "grad_norm": 943.1874389648438,
      "learning_rate": 4.928904111789805e-05,
      "loss": 122.9224,
      "step": 20850
    },
    {
      "epoch": 0.16855339813670117,
      "grad_norm": 582.3085327148438,
      "learning_rate": 4.9287370370377106e-05,
      "loss": 154.6284,
      "step": 20860
    },
    {
      "epoch": 0.1686342003409853,
      "grad_norm": 363.685546875,
      "learning_rate": 4.928569769042156e-05,
      "loss": 159.0539,
      "step": 20870
    },
    {
      "epoch": 0.16871500254526944,
      "grad_norm": 1049.855224609375,
      "learning_rate": 4.928402307816451e-05,
      "loss": 140.1967,
      "step": 20880
    },
    {
      "epoch": 0.16879580474955358,
      "grad_norm": 912.501708984375,
      "learning_rate": 4.92823465337392e-05,
      "loss": 124.3262,
      "step": 20890
    },
    {
      "epoch": 0.1688766069538377,
      "grad_norm": 869.2931518554688,
      "learning_rate": 4.9280668057279014e-05,
      "loss": 125.99,
      "step": 20900
    },
    {
      "epoch": 0.16895740915812182,
      "grad_norm": 834.9353637695312,
      "learning_rate": 4.9278987648917495e-05,
      "loss": 121.5135,
      "step": 20910
    },
    {
      "epoch": 0.16903821136240596,
      "grad_norm": 933.3566284179688,
      "learning_rate": 4.9277305308788365e-05,
      "loss": 135.7604,
      "step": 20920
    },
    {
      "epoch": 0.1691190135666901,
      "grad_norm": 6561.35693359375,
      "learning_rate": 4.927562103702547e-05,
      "loss": 169.7086,
      "step": 20930
    },
    {
      "epoch": 0.16919981577097423,
      "grad_norm": 1966.267822265625,
      "learning_rate": 4.92739348337628e-05,
      "loss": 135.3359,
      "step": 20940
    },
    {
      "epoch": 0.16928061797525837,
      "grad_norm": 1062.4327392578125,
      "learning_rate": 4.927224669913456e-05,
      "loss": 130.8302,
      "step": 20950
    },
    {
      "epoch": 0.1693614201795425,
      "grad_norm": 1002.7949829101562,
      "learning_rate": 4.927055663327503e-05,
      "loss": 152.034,
      "step": 20960
    },
    {
      "epoch": 0.16944222238382664,
      "grad_norm": 1231.685302734375,
      "learning_rate": 4.92688646363187e-05,
      "loss": 125.2035,
      "step": 20970
    },
    {
      "epoch": 0.16952302458811078,
      "grad_norm": 1325.4017333984375,
      "learning_rate": 4.9267170708400196e-05,
      "loss": 153.9901,
      "step": 20980
    },
    {
      "epoch": 0.16960382679239489,
      "grad_norm": 1044.0921630859375,
      "learning_rate": 4.9265474849654284e-05,
      "loss": 160.7131,
      "step": 20990
    },
    {
      "epoch": 0.16968462899667902,
      "grad_norm": 706.5172119140625,
      "learning_rate": 4.92637770602159e-05,
      "loss": 150.8885,
      "step": 21000
    },
    {
      "epoch": 0.16976543120096316,
      "grad_norm": 443.4945373535156,
      "learning_rate": 4.9262077340220135e-05,
      "loss": 137.6814,
      "step": 21010
    },
    {
      "epoch": 0.1698462334052473,
      "grad_norm": 3524.959228515625,
      "learning_rate": 4.926037568980223e-05,
      "loss": 131.1957,
      "step": 21020
    },
    {
      "epoch": 0.16992703560953143,
      "grad_norm": 1061.660888671875,
      "learning_rate": 4.925867210909756e-05,
      "loss": 109.0972,
      "step": 21030
    },
    {
      "epoch": 0.17000783781381557,
      "grad_norm": 977.8228759765625,
      "learning_rate": 4.925696659824169e-05,
      "loss": 140.1199,
      "step": 21040
    },
    {
      "epoch": 0.1700886400180997,
      "grad_norm": 886.8487548828125,
      "learning_rate": 4.925525915737031e-05,
      "loss": 123.5597,
      "step": 21050
    },
    {
      "epoch": 0.17016944222238384,
      "grad_norm": 2220.47314453125,
      "learning_rate": 4.925354978661928e-05,
      "loss": 116.6628,
      "step": 21060
    },
    {
      "epoch": 0.17025024442666795,
      "grad_norm": 750.9844970703125,
      "learning_rate": 4.925183848612459e-05,
      "loss": 150.5052,
      "step": 21070
    },
    {
      "epoch": 0.17033104663095208,
      "grad_norm": 1006.42333984375,
      "learning_rate": 4.9250125256022416e-05,
      "loss": 119.5853,
      "step": 21080
    },
    {
      "epoch": 0.17041184883523622,
      "grad_norm": 1162.7249755859375,
      "learning_rate": 4.9248410096449075e-05,
      "loss": 137.6988,
      "step": 21090
    },
    {
      "epoch": 0.17049265103952035,
      "grad_norm": 666.7645263671875,
      "learning_rate": 4.9246693007541024e-05,
      "loss": 181.7554,
      "step": 21100
    },
    {
      "epoch": 0.1705734532438045,
      "grad_norm": 1177.0545654296875,
      "learning_rate": 4.9244973989434886e-05,
      "loss": 157.994,
      "step": 21110
    },
    {
      "epoch": 0.17065425544808863,
      "grad_norm": 1034.1756591796875,
      "learning_rate": 4.924325304226745e-05,
      "loss": 135.4122,
      "step": 21120
    },
    {
      "epoch": 0.17073505765237276,
      "grad_norm": 1536.7457275390625,
      "learning_rate": 4.9241530166175614e-05,
      "loss": 153.4128,
      "step": 21130
    },
    {
      "epoch": 0.1708158598566569,
      "grad_norm": 881.2988891601562,
      "learning_rate": 4.9239805361296485e-05,
      "loss": 161.9859,
      "step": 21140
    },
    {
      "epoch": 0.17089666206094103,
      "grad_norm": 1522.52197265625,
      "learning_rate": 4.923807862776728e-05,
      "loss": 178.6479,
      "step": 21150
    },
    {
      "epoch": 0.17097746426522514,
      "grad_norm": 658.5277709960938,
      "learning_rate": 4.9236349965725406e-05,
      "loss": 130.6969,
      "step": 21160
    },
    {
      "epoch": 0.17105826646950928,
      "grad_norm": 865.1202392578125,
      "learning_rate": 4.9234619375308396e-05,
      "loss": 125.0552,
      "step": 21170
    },
    {
      "epoch": 0.17113906867379342,
      "grad_norm": 1105.9573974609375,
      "learning_rate": 4.9232886856653936e-05,
      "loss": 148.4609,
      "step": 21180
    },
    {
      "epoch": 0.17121987087807755,
      "grad_norm": 762.8492431640625,
      "learning_rate": 4.923115240989989e-05,
      "loss": 177.1173,
      "step": 21190
    },
    {
      "epoch": 0.1713006730823617,
      "grad_norm": 977.0111083984375,
      "learning_rate": 4.922941603518425e-05,
      "loss": 131.6727,
      "step": 21200
    },
    {
      "epoch": 0.17138147528664582,
      "grad_norm": 864.1790771484375,
      "learning_rate": 4.922767773264517e-05,
      "loss": 153.2473,
      "step": 21210
    },
    {
      "epoch": 0.17146227749092996,
      "grad_norm": 576.5042724609375,
      "learning_rate": 4.9225937502420974e-05,
      "loss": 142.9099,
      "step": 21220
    },
    {
      "epoch": 0.1715430796952141,
      "grad_norm": 826.1639404296875,
      "learning_rate": 4.9224195344650105e-05,
      "loss": 165.9666,
      "step": 21230
    },
    {
      "epoch": 0.17162388189949823,
      "grad_norm": 1021.5435180664062,
      "learning_rate": 4.922245125947119e-05,
      "loss": 131.2517,
      "step": 21240
    },
    {
      "epoch": 0.17170468410378234,
      "grad_norm": 750.6788330078125,
      "learning_rate": 4.9220705247022985e-05,
      "loss": 128.4156,
      "step": 21250
    },
    {
      "epoch": 0.17178548630806648,
      "grad_norm": 1202.5364990234375,
      "learning_rate": 4.921895730744443e-05,
      "loss": 165.3158,
      "step": 21260
    },
    {
      "epoch": 0.1718662885123506,
      "grad_norm": 695.836181640625,
      "learning_rate": 4.921720744087459e-05,
      "loss": 108.6388,
      "step": 21270
    },
    {
      "epoch": 0.17194709071663475,
      "grad_norm": 658.8826904296875,
      "learning_rate": 4.921545564745269e-05,
      "loss": 134.6856,
      "step": 21280
    },
    {
      "epoch": 0.17202789292091888,
      "grad_norm": 2134.118896484375,
      "learning_rate": 4.9213701927318134e-05,
      "loss": 147.9883,
      "step": 21290
    },
    {
      "epoch": 0.17210869512520302,
      "grad_norm": 981.736083984375,
      "learning_rate": 4.921194628061043e-05,
      "loss": 134.1546,
      "step": 21300
    },
    {
      "epoch": 0.17218949732948716,
      "grad_norm": 619.3984985351562,
      "learning_rate": 4.921018870746928e-05,
      "loss": 119.4398,
      "step": 21310
    },
    {
      "epoch": 0.1722702995337713,
      "grad_norm": 710.1690673828125,
      "learning_rate": 4.9208429208034525e-05,
      "loss": 145.9413,
      "step": 21320
    },
    {
      "epoch": 0.1723511017380554,
      "grad_norm": 1467.351806640625,
      "learning_rate": 4.920666778244616e-05,
      "loss": 159.9146,
      "step": 21330
    },
    {
      "epoch": 0.17243190394233954,
      "grad_norm": 862.062255859375,
      "learning_rate": 4.920490443084433e-05,
      "loss": 129.9643,
      "step": 21340
    },
    {
      "epoch": 0.17251270614662367,
      "grad_norm": 1797.9549560546875,
      "learning_rate": 4.920313915336934e-05,
      "loss": 144.6173,
      "step": 21350
    },
    {
      "epoch": 0.1725935083509078,
      "grad_norm": 1055.6019287109375,
      "learning_rate": 4.920137195016165e-05,
      "loss": 156.1974,
      "step": 21360
    },
    {
      "epoch": 0.17267431055519195,
      "grad_norm": 8324.7998046875,
      "learning_rate": 4.919960282136185e-05,
      "loss": 164.71,
      "step": 21370
    },
    {
      "epoch": 0.17275511275947608,
      "grad_norm": 971.8815307617188,
      "learning_rate": 4.919783176711074e-05,
      "loss": 153.0036,
      "step": 21380
    },
    {
      "epoch": 0.17283591496376022,
      "grad_norm": 988.0467529296875,
      "learning_rate": 4.9196058787549184e-05,
      "loss": 152.4217,
      "step": 21390
    },
    {
      "epoch": 0.17291671716804435,
      "grad_norm": 570.156494140625,
      "learning_rate": 4.919428388281829e-05,
      "loss": 127.8879,
      "step": 21400
    },
    {
      "epoch": 0.1729975193723285,
      "grad_norm": 1578.5853271484375,
      "learning_rate": 4.9192507053059255e-05,
      "loss": 148.3654,
      "step": 21410
    },
    {
      "epoch": 0.1730783215766126,
      "grad_norm": 1456.0301513671875,
      "learning_rate": 4.919072829841347e-05,
      "loss": 178.0747,
      "step": 21420
    },
    {
      "epoch": 0.17315912378089673,
      "grad_norm": 588.3489990234375,
      "learning_rate": 4.918894761902245e-05,
      "loss": 143.8035,
      "step": 21430
    },
    {
      "epoch": 0.17323992598518087,
      "grad_norm": 927.64404296875,
      "learning_rate": 4.918716501502789e-05,
      "loss": 126.7022,
      "step": 21440
    },
    {
      "epoch": 0.173320728189465,
      "grad_norm": 1143.9464111328125,
      "learning_rate": 4.9185380486571595e-05,
      "loss": 79.5578,
      "step": 21450
    },
    {
      "epoch": 0.17340153039374914,
      "grad_norm": 838.281494140625,
      "learning_rate": 4.918359403379559e-05,
      "loss": 177.1575,
      "step": 21460
    },
    {
      "epoch": 0.17348233259803328,
      "grad_norm": 1176.1307373046875,
      "learning_rate": 4.918180565684198e-05,
      "loss": 151.7111,
      "step": 21470
    },
    {
      "epoch": 0.17356313480231741,
      "grad_norm": 1099.8289794921875,
      "learning_rate": 4.9180015355853084e-05,
      "loss": 152.3932,
      "step": 21480
    },
    {
      "epoch": 0.17364393700660155,
      "grad_norm": 1128.26123046875,
      "learning_rate": 4.917822313097134e-05,
      "loss": 116.6667,
      "step": 21490
    },
    {
      "epoch": 0.17372473921088566,
      "grad_norm": 563.2874755859375,
      "learning_rate": 4.917642898233933e-05,
      "loss": 142.594,
      "step": 21500
    },
    {
      "epoch": 0.1738055414151698,
      "grad_norm": 1359.58984375,
      "learning_rate": 4.917463291009983e-05,
      "loss": 133.1546,
      "step": 21510
    },
    {
      "epoch": 0.17388634361945393,
      "grad_norm": 848.3027954101562,
      "learning_rate": 4.917283491439574e-05,
      "loss": 152.585,
      "step": 21520
    },
    {
      "epoch": 0.17396714582373807,
      "grad_norm": 857.1946411132812,
      "learning_rate": 4.91710349953701e-05,
      "loss": 178.7701,
      "step": 21530
    },
    {
      "epoch": 0.1740479480280222,
      "grad_norm": 1514.6451416015625,
      "learning_rate": 4.916923315316615e-05,
      "loss": 165.9126,
      "step": 21540
    },
    {
      "epoch": 0.17412875023230634,
      "grad_norm": 1154.829833984375,
      "learning_rate": 4.916742938792723e-05,
      "loss": 129.4643,
      "step": 21550
    },
    {
      "epoch": 0.17420955243659048,
      "grad_norm": 1806.4046630859375,
      "learning_rate": 4.9165623699796875e-05,
      "loss": 131.133,
      "step": 21560
    },
    {
      "epoch": 0.1742903546408746,
      "grad_norm": 2297.58642578125,
      "learning_rate": 4.916381608891874e-05,
      "loss": 181.9885,
      "step": 21570
    },
    {
      "epoch": 0.17437115684515875,
      "grad_norm": 889.8167724609375,
      "learning_rate": 4.916200655543667e-05,
      "loss": 124.0278,
      "step": 21580
    },
    {
      "epoch": 0.17445195904944286,
      "grad_norm": 652.2991333007812,
      "learning_rate": 4.916019509949461e-05,
      "loss": 129.1368,
      "step": 21590
    },
    {
      "epoch": 0.174532761253727,
      "grad_norm": 2002.193603515625,
      "learning_rate": 4.915838172123671e-05,
      "loss": 156.8969,
      "step": 21600
    },
    {
      "epoch": 0.17461356345801113,
      "grad_norm": 1632.37646484375,
      "learning_rate": 4.915656642080726e-05,
      "loss": 134.3347,
      "step": 21610
    },
    {
      "epoch": 0.17469436566229526,
      "grad_norm": 1093.335205078125,
      "learning_rate": 4.915474919835067e-05,
      "loss": 134.8566,
      "step": 21620
    },
    {
      "epoch": 0.1747751678665794,
      "grad_norm": 1129.8870849609375,
      "learning_rate": 4.915293005401155e-05,
      "loss": 140.7855,
      "step": 21630
    },
    {
      "epoch": 0.17485597007086354,
      "grad_norm": 1071.5859375,
      "learning_rate": 4.915110898793462e-05,
      "loss": 164.049,
      "step": 21640
    },
    {
      "epoch": 0.17493677227514767,
      "grad_norm": 729.1155395507812,
      "learning_rate": 4.9149286000264805e-05,
      "loss": 113.2319,
      "step": 21650
    },
    {
      "epoch": 0.1750175744794318,
      "grad_norm": 753.5103149414062,
      "learning_rate": 4.9147461091147125e-05,
      "loss": 102.7337,
      "step": 21660
    },
    {
      "epoch": 0.17509837668371594,
      "grad_norm": 531.7659301757812,
      "learning_rate": 4.914563426072678e-05,
      "loss": 149.2153,
      "step": 21670
    },
    {
      "epoch": 0.17517917888800005,
      "grad_norm": 1287.9884033203125,
      "learning_rate": 4.914380550914914e-05,
      "loss": 146.6886,
      "step": 21680
    },
    {
      "epoch": 0.1752599810922842,
      "grad_norm": 1483.9490966796875,
      "learning_rate": 4.9141974836559704e-05,
      "loss": 157.0422,
      "step": 21690
    },
    {
      "epoch": 0.17534078329656833,
      "grad_norm": 899.7554931640625,
      "learning_rate": 4.9140142243104116e-05,
      "loss": 149.8844,
      "step": 21700
    },
    {
      "epoch": 0.17542158550085246,
      "grad_norm": 1013.6568603515625,
      "learning_rate": 4.91383077289282e-05,
      "loss": 99.9814,
      "step": 21710
    },
    {
      "epoch": 0.1755023877051366,
      "grad_norm": 1174.3260498046875,
      "learning_rate": 4.9136471294177924e-05,
      "loss": 138.6684,
      "step": 21720
    },
    {
      "epoch": 0.17558318990942073,
      "grad_norm": 1020.420166015625,
      "learning_rate": 4.913463293899939e-05,
      "loss": 165.4781,
      "step": 21730
    },
    {
      "epoch": 0.17566399211370487,
      "grad_norm": 1076.771728515625,
      "learning_rate": 4.913279266353888e-05,
      "loss": 138.3935,
      "step": 21740
    },
    {
      "epoch": 0.175744794317989,
      "grad_norm": 1392.3963623046875,
      "learning_rate": 4.9130950467942814e-05,
      "loss": 130.0977,
      "step": 21750
    },
    {
      "epoch": 0.17582559652227311,
      "grad_norm": 593.2973022460938,
      "learning_rate": 4.912910635235777e-05,
      "loss": 116.5755,
      "step": 21760
    },
    {
      "epoch": 0.17590639872655725,
      "grad_norm": 1103.0938720703125,
      "learning_rate": 4.9127260316930466e-05,
      "loss": 138.2039,
      "step": 21770
    },
    {
      "epoch": 0.1759872009308414,
      "grad_norm": 737.562744140625,
      "learning_rate": 4.912541236180779e-05,
      "loss": 122.1219,
      "step": 21780
    },
    {
      "epoch": 0.17606800313512552,
      "grad_norm": 860.9588623046875,
      "learning_rate": 4.9123562487136774e-05,
      "loss": 153.3838,
      "step": 21790
    },
    {
      "epoch": 0.17614880533940966,
      "grad_norm": 1145.1170654296875,
      "learning_rate": 4.912171069306459e-05,
      "loss": 123.9352,
      "step": 21800
    },
    {
      "epoch": 0.1762296075436938,
      "grad_norm": 940.3277587890625,
      "learning_rate": 4.911985697973861e-05,
      "loss": 142.1018,
      "step": 21810
    },
    {
      "epoch": 0.17631040974797793,
      "grad_norm": 815.0631713867188,
      "learning_rate": 4.911800134730629e-05,
      "loss": 151.9103,
      "step": 21820
    },
    {
      "epoch": 0.17639121195226207,
      "grad_norm": 1341.9578857421875,
      "learning_rate": 4.9116143795915295e-05,
      "loss": 146.6557,
      "step": 21830
    },
    {
      "epoch": 0.1764720141565462,
      "grad_norm": 861.9169311523438,
      "learning_rate": 4.9114284325713416e-05,
      "loss": 131.5514,
      "step": 21840
    },
    {
      "epoch": 0.1765528163608303,
      "grad_norm": 2296.96484375,
      "learning_rate": 4.91124229368486e-05,
      "loss": 127.922,
      "step": 21850
    },
    {
      "epoch": 0.17663361856511445,
      "grad_norm": 3126.313232421875,
      "learning_rate": 4.911055962946896e-05,
      "loss": 178.3461,
      "step": 21860
    },
    {
      "epoch": 0.17671442076939858,
      "grad_norm": 1112.87646484375,
      "learning_rate": 4.910869440372274e-05,
      "loss": 131.921,
      "step": 21870
    },
    {
      "epoch": 0.17679522297368272,
      "grad_norm": 1277.1387939453125,
      "learning_rate": 4.910682725975835e-05,
      "loss": 172.0795,
      "step": 21880
    },
    {
      "epoch": 0.17687602517796686,
      "grad_norm": 1765.8804931640625,
      "learning_rate": 4.910495819772434e-05,
      "loss": 116.7105,
      "step": 21890
    },
    {
      "epoch": 0.176956827382251,
      "grad_norm": 1198.19384765625,
      "learning_rate": 4.9103087217769445e-05,
      "loss": 145.228,
      "step": 21900
    },
    {
      "epoch": 0.17703762958653513,
      "grad_norm": 926.2941284179688,
      "learning_rate": 4.910121432004252e-05,
      "loss": 161.9553,
      "step": 21910
    },
    {
      "epoch": 0.17711843179081926,
      "grad_norm": 759.7756958007812,
      "learning_rate": 4.9099339504692573e-05,
      "loss": 113.5963,
      "step": 21920
    },
    {
      "epoch": 0.1771992339951034,
      "grad_norm": 816.7711181640625,
      "learning_rate": 4.909746277186879e-05,
      "loss": 145.3244,
      "step": 21930
    },
    {
      "epoch": 0.1772800361993875,
      "grad_norm": 1038.8468017578125,
      "learning_rate": 4.909558412172047e-05,
      "loss": 175.112,
      "step": 21940
    },
    {
      "epoch": 0.17736083840367164,
      "grad_norm": 1387.3388671875,
      "learning_rate": 4.909370355439712e-05,
      "loss": 124.688,
      "step": 21950
    },
    {
      "epoch": 0.17744164060795578,
      "grad_norm": 666.5802612304688,
      "learning_rate": 4.909182107004835e-05,
      "loss": 85.6291,
      "step": 21960
    },
    {
      "epoch": 0.17752244281223992,
      "grad_norm": 1517.434326171875,
      "learning_rate": 4.908993666882395e-05,
      "loss": 125.7463,
      "step": 21970
    },
    {
      "epoch": 0.17760324501652405,
      "grad_norm": 746.6646118164062,
      "learning_rate": 4.9088050350873835e-05,
      "loss": 143.7797,
      "step": 21980
    },
    {
      "epoch": 0.1776840472208082,
      "grad_norm": 838.4737548828125,
      "learning_rate": 4.9086162116348114e-05,
      "loss": 120.7253,
      "step": 21990
    },
    {
      "epoch": 0.17776484942509233,
      "grad_norm": 1266.503662109375,
      "learning_rate": 4.9084271965397014e-05,
      "loss": 141.1467,
      "step": 22000
    },
    {
      "epoch": 0.17784565162937646,
      "grad_norm": 1920.794677734375,
      "learning_rate": 4.9082379898170914e-05,
      "loss": 143.4987,
      "step": 22010
    },
    {
      "epoch": 0.17792645383366057,
      "grad_norm": 1311.13720703125,
      "learning_rate": 4.908048591482038e-05,
      "loss": 152.7263,
      "step": 22020
    },
    {
      "epoch": 0.1780072560379447,
      "grad_norm": 1175.812744140625,
      "learning_rate": 4.9078590015496096e-05,
      "loss": 129.1362,
      "step": 22030
    },
    {
      "epoch": 0.17808805824222884,
      "grad_norm": 1120.2232666015625,
      "learning_rate": 4.907669220034891e-05,
      "loss": 138.7439,
      "step": 22040
    },
    {
      "epoch": 0.17816886044651298,
      "grad_norm": 1485.8729248046875,
      "learning_rate": 4.9074792469529815e-05,
      "loss": 274.2296,
      "step": 22050
    },
    {
      "epoch": 0.1782496626507971,
      "grad_norm": 969.6702270507812,
      "learning_rate": 4.9072890823189974e-05,
      "loss": 119.1664,
      "step": 22060
    },
    {
      "epoch": 0.17833046485508125,
      "grad_norm": 941.8471069335938,
      "learning_rate": 4.907098726148069e-05,
      "loss": 160.9961,
      "step": 22070
    },
    {
      "epoch": 0.17841126705936539,
      "grad_norm": 1140.767333984375,
      "learning_rate": 4.906908178455342e-05,
      "loss": 145.0228,
      "step": 22080
    },
    {
      "epoch": 0.17849206926364952,
      "grad_norm": 715.76904296875,
      "learning_rate": 4.9067174392559776e-05,
      "loss": 103.8144,
      "step": 22090
    },
    {
      "epoch": 0.17857287146793366,
      "grad_norm": 1049.420654296875,
      "learning_rate": 4.906526508565152e-05,
      "loss": 124.3415,
      "step": 22100
    },
    {
      "epoch": 0.17865367367221777,
      "grad_norm": 1215.5096435546875,
      "learning_rate": 4.9063353863980565e-05,
      "loss": 182.3524,
      "step": 22110
    },
    {
      "epoch": 0.1787344758765019,
      "grad_norm": 1063.1181640625,
      "learning_rate": 4.906144072769898e-05,
      "loss": 122.475,
      "step": 22120
    },
    {
      "epoch": 0.17881527808078604,
      "grad_norm": 1125.4080810546875,
      "learning_rate": 4.9059525676958986e-05,
      "loss": 139.1991,
      "step": 22130
    },
    {
      "epoch": 0.17889608028507017,
      "grad_norm": 1022.3780517578125,
      "learning_rate": 4.905760871191295e-05,
      "loss": 195.8249,
      "step": 22140
    },
    {
      "epoch": 0.1789768824893543,
      "grad_norm": 1560.5745849609375,
      "learning_rate": 4.9055689832713396e-05,
      "loss": 125.5297,
      "step": 22150
    },
    {
      "epoch": 0.17905768469363845,
      "grad_norm": 5475.33447265625,
      "learning_rate": 4.9053769039513006e-05,
      "loss": 148.3278,
      "step": 22160
    },
    {
      "epoch": 0.17913848689792258,
      "grad_norm": 737.4940795898438,
      "learning_rate": 4.90518463324646e-05,
      "loss": 140.7979,
      "step": 22170
    },
    {
      "epoch": 0.17921928910220672,
      "grad_norm": 627.595458984375,
      "learning_rate": 4.904992171172117e-05,
      "loss": 157.4451,
      "step": 22180
    },
    {
      "epoch": 0.17930009130649083,
      "grad_norm": 734.084228515625,
      "learning_rate": 4.904799517743585e-05,
      "loss": 139.5384,
      "step": 22190
    },
    {
      "epoch": 0.17938089351077496,
      "grad_norm": 1306.1624755859375,
      "learning_rate": 4.9046066729761905e-05,
      "loss": 118.6943,
      "step": 22200
    },
    {
      "epoch": 0.1794616957150591,
      "grad_norm": 1220.8062744140625,
      "learning_rate": 4.904413636885279e-05,
      "loss": 168.1177,
      "step": 22210
    },
    {
      "epoch": 0.17954249791934324,
      "grad_norm": 1183.083251953125,
      "learning_rate": 4.90422040948621e-05,
      "loss": 158.66,
      "step": 22220
    },
    {
      "epoch": 0.17962330012362737,
      "grad_norm": 1026.29345703125,
      "learning_rate": 4.904026990794356e-05,
      "loss": 128.3199,
      "step": 22230
    },
    {
      "epoch": 0.1797041023279115,
      "grad_norm": 771.4765625,
      "learning_rate": 4.9038333808251074e-05,
      "loss": 111.4788,
      "step": 22240
    },
    {
      "epoch": 0.17978490453219564,
      "grad_norm": 1059.020751953125,
      "learning_rate": 4.90363957959387e-05,
      "loss": 134.2409,
      "step": 22250
    },
    {
      "epoch": 0.17986570673647978,
      "grad_norm": 1103.791748046875,
      "learning_rate": 4.903445587116061e-05,
      "loss": 170.8439,
      "step": 22260
    },
    {
      "epoch": 0.17994650894076392,
      "grad_norm": 622.0547485351562,
      "learning_rate": 4.9032514034071175e-05,
      "loss": 122.9244,
      "step": 22270
    },
    {
      "epoch": 0.18002731114504802,
      "grad_norm": 958.7100830078125,
      "learning_rate": 4.903057028482489e-05,
      "loss": 99.9289,
      "step": 22280
    },
    {
      "epoch": 0.18010811334933216,
      "grad_norm": 996.9059448242188,
      "learning_rate": 4.902862462357641e-05,
      "loss": 164.1096,
      "step": 22290
    },
    {
      "epoch": 0.1801889155536163,
      "grad_norm": 481.1267395019531,
      "learning_rate": 4.9026677050480555e-05,
      "loss": 118.1814,
      "step": 22300
    },
    {
      "epoch": 0.18026971775790043,
      "grad_norm": 1068.948486328125,
      "learning_rate": 4.9024727565692274e-05,
      "loss": 144.3041,
      "step": 22310
    },
    {
      "epoch": 0.18035051996218457,
      "grad_norm": 997.0437622070312,
      "learning_rate": 4.9022776169366666e-05,
      "loss": 155.9985,
      "step": 22320
    },
    {
      "epoch": 0.1804313221664687,
      "grad_norm": 1654.6068115234375,
      "learning_rate": 4.902082286165902e-05,
      "loss": 164.7646,
      "step": 22330
    },
    {
      "epoch": 0.18051212437075284,
      "grad_norm": 3154.902099609375,
      "learning_rate": 4.901886764272474e-05,
      "loss": 98.5395,
      "step": 22340
    },
    {
      "epoch": 0.18059292657503698,
      "grad_norm": 2299.2890625,
      "learning_rate": 4.901691051271939e-05,
      "loss": 146.594,
      "step": 22350
    },
    {
      "epoch": 0.1806737287793211,
      "grad_norm": 1197.87451171875,
      "learning_rate": 4.90149514717987e-05,
      "loss": 145.7994,
      "step": 22360
    },
    {
      "epoch": 0.18075453098360522,
      "grad_norm": 1042.1507568359375,
      "learning_rate": 4.901299052011852e-05,
      "loss": 144.4632,
      "step": 22370
    },
    {
      "epoch": 0.18083533318788936,
      "grad_norm": 1308.17041015625,
      "learning_rate": 4.90110276578349e-05,
      "loss": 119.5589,
      "step": 22380
    },
    {
      "epoch": 0.1809161353921735,
      "grad_norm": 904.1051025390625,
      "learning_rate": 4.900906288510401e-05,
      "loss": 116.1322,
      "step": 22390
    },
    {
      "epoch": 0.18099693759645763,
      "grad_norm": 807.6878051757812,
      "learning_rate": 4.900709620208216e-05,
      "loss": 179.056,
      "step": 22400
    },
    {
      "epoch": 0.18107773980074177,
      "grad_norm": 644.4304809570312,
      "learning_rate": 4.900512760892585e-05,
      "loss": 110.5187,
      "step": 22410
    },
    {
      "epoch": 0.1811585420050259,
      "grad_norm": 924.395263671875,
      "learning_rate": 4.9003157105791706e-05,
      "loss": 116.3889,
      "step": 22420
    },
    {
      "epoch": 0.18123934420931004,
      "grad_norm": 1086.564208984375,
      "learning_rate": 4.9001184692836505e-05,
      "loss": 121.7575,
      "step": 22430
    },
    {
      "epoch": 0.18132014641359417,
      "grad_norm": 876.027587890625,
      "learning_rate": 4.8999210370217194e-05,
      "loss": 159.8317,
      "step": 22440
    },
    {
      "epoch": 0.18140094861787828,
      "grad_norm": 1000.0255126953125,
      "learning_rate": 4.899723413809085e-05,
      "loss": 139.3787,
      "step": 22450
    },
    {
      "epoch": 0.18148175082216242,
      "grad_norm": 2290.170654296875,
      "learning_rate": 4.899525599661472e-05,
      "loss": 97.0173,
      "step": 22460
    },
    {
      "epoch": 0.18156255302644655,
      "grad_norm": 1271.3829345703125,
      "learning_rate": 4.899327594594619e-05,
      "loss": 131.2777,
      "step": 22470
    },
    {
      "epoch": 0.1816433552307307,
      "grad_norm": 907.057861328125,
      "learning_rate": 4.899129398624281e-05,
      "loss": 101.1429,
      "step": 22480
    },
    {
      "epoch": 0.18172415743501483,
      "grad_norm": 1061.6846923828125,
      "learning_rate": 4.898931011766228e-05,
      "loss": 107.645,
      "step": 22490
    },
    {
      "epoch": 0.18180495963929896,
      "grad_norm": 1035.3082275390625,
      "learning_rate": 4.898732434036244e-05,
      "loss": 121.6947,
      "step": 22500
    },
    {
      "epoch": 0.1818857618435831,
      "grad_norm": 1202.3555908203125,
      "learning_rate": 4.898533665450128e-05,
      "loss": 162.0105,
      "step": 22510
    },
    {
      "epoch": 0.18196656404786724,
      "grad_norm": 1121.3282470703125,
      "learning_rate": 4.898334706023696e-05,
      "loss": 113.9376,
      "step": 22520
    },
    {
      "epoch": 0.18204736625215137,
      "grad_norm": 1393.017578125,
      "learning_rate": 4.8981355557727796e-05,
      "loss": 139.8284,
      "step": 22530
    },
    {
      "epoch": 0.18212816845643548,
      "grad_norm": 2622.062744140625,
      "learning_rate": 4.897936214713222e-05,
      "loss": 168.307,
      "step": 22540
    },
    {
      "epoch": 0.18220897066071962,
      "grad_norm": 804.2191772460938,
      "learning_rate": 4.897736682860885e-05,
      "loss": 146.0737,
      "step": 22550
    },
    {
      "epoch": 0.18228977286500375,
      "grad_norm": 1040.8131103515625,
      "learning_rate": 4.897536960231644e-05,
      "loss": 157.3722,
      "step": 22560
    },
    {
      "epoch": 0.1823705750692879,
      "grad_norm": 677.9693603515625,
      "learning_rate": 4.89733704684139e-05,
      "loss": 103.1337,
      "step": 22570
    },
    {
      "epoch": 0.18245137727357202,
      "grad_norm": 1318.36376953125,
      "learning_rate": 4.89713694270603e-05,
      "loss": 141.4517,
      "step": 22580
    },
    {
      "epoch": 0.18253217947785616,
      "grad_norm": 1170.07861328125,
      "learning_rate": 4.8969366478414854e-05,
      "loss": 108.5731,
      "step": 22590
    },
    {
      "epoch": 0.1826129816821403,
      "grad_norm": 1243.060791015625,
      "learning_rate": 4.896736162263691e-05,
      "loss": 173.4797,
      "step": 22600
    },
    {
      "epoch": 0.18269378388642443,
      "grad_norm": 624.64111328125,
      "learning_rate": 4.8965354859886006e-05,
      "loss": 103.4098,
      "step": 22610
    },
    {
      "epoch": 0.18277458609070854,
      "grad_norm": 982.3737182617188,
      "learning_rate": 4.89633461903218e-05,
      "loss": 122.2358,
      "step": 22620
    },
    {
      "epoch": 0.18285538829499268,
      "grad_norm": 1259.17578125,
      "learning_rate": 4.8961335614104115e-05,
      "loss": 124.4892,
      "step": 22630
    },
    {
      "epoch": 0.1829361904992768,
      "grad_norm": 1049.9534912109375,
      "learning_rate": 4.895932313139292e-05,
      "loss": 165.0955,
      "step": 22640
    },
    {
      "epoch": 0.18301699270356095,
      "grad_norm": 805.4934692382812,
      "learning_rate": 4.895730874234834e-05,
      "loss": 114.1204,
      "step": 22650
    },
    {
      "epoch": 0.18309779490784509,
      "grad_norm": 1224.6595458984375,
      "learning_rate": 4.895529244713066e-05,
      "loss": 115.779,
      "step": 22660
    },
    {
      "epoch": 0.18317859711212922,
      "grad_norm": 1112.3597412109375,
      "learning_rate": 4.895327424590029e-05,
      "loss": 130.0822,
      "step": 22670
    },
    {
      "epoch": 0.18325939931641336,
      "grad_norm": 1353.516845703125,
      "learning_rate": 4.895125413881783e-05,
      "loss": 171.6577,
      "step": 22680
    },
    {
      "epoch": 0.1833402015206975,
      "grad_norm": 1338.9501953125,
      "learning_rate": 4.8949232126044e-05,
      "loss": 109.7441,
      "step": 22690
    },
    {
      "epoch": 0.18342100372498163,
      "grad_norm": 1896.7581787109375,
      "learning_rate": 4.894720820773967e-05,
      "loss": 174.3409,
      "step": 22700
    },
    {
      "epoch": 0.18350180592926574,
      "grad_norm": 766.3515014648438,
      "learning_rate": 4.894518238406589e-05,
      "loss": 137.257,
      "step": 22710
    },
    {
      "epoch": 0.18358260813354987,
      "grad_norm": 691.5870971679688,
      "learning_rate": 4.8943154655183844e-05,
      "loss": 100.2154,
      "step": 22720
    },
    {
      "epoch": 0.183663410337834,
      "grad_norm": 5152.01025390625,
      "learning_rate": 4.894112502125487e-05,
      "loss": 173.3306,
      "step": 22730
    },
    {
      "epoch": 0.18374421254211815,
      "grad_norm": 788.1614379882812,
      "learning_rate": 4.8939093482440444e-05,
      "loss": 142.851,
      "step": 22740
    },
    {
      "epoch": 0.18382501474640228,
      "grad_norm": 1022.0597534179688,
      "learning_rate": 4.8937060038902224e-05,
      "loss": 160.1116,
      "step": 22750
    },
    {
      "epoch": 0.18390581695068642,
      "grad_norm": 757.3877563476562,
      "learning_rate": 4.8935024690801986e-05,
      "loss": 104.9073,
      "step": 22760
    },
    {
      "epoch": 0.18398661915497055,
      "grad_norm": 1713.2584228515625,
      "learning_rate": 4.893298743830168e-05,
      "loss": 141.8529,
      "step": 22770
    },
    {
      "epoch": 0.1840674213592547,
      "grad_norm": 1372.67822265625,
      "learning_rate": 4.89309482815634e-05,
      "loss": 157.4392,
      "step": 22780
    },
    {
      "epoch": 0.18414822356353883,
      "grad_norm": 900.3597412109375,
      "learning_rate": 4.892890722074941e-05,
      "loss": 161.8838,
      "step": 22790
    },
    {
      "epoch": 0.18422902576782293,
      "grad_norm": 1206.832763671875,
      "learning_rate": 4.8926864256022074e-05,
      "loss": 142.6541,
      "step": 22800
    },
    {
      "epoch": 0.18430982797210707,
      "grad_norm": 3348.473388671875,
      "learning_rate": 4.892481938754396e-05,
      "loss": 153.2709,
      "step": 22810
    },
    {
      "epoch": 0.1843906301763912,
      "grad_norm": 1615.8541259765625,
      "learning_rate": 4.892277261547778e-05,
      "loss": 176.116,
      "step": 22820
    },
    {
      "epoch": 0.18447143238067534,
      "grad_norm": 1190.1331787109375,
      "learning_rate": 4.892072393998636e-05,
      "loss": 130.4555,
      "step": 22830
    },
    {
      "epoch": 0.18455223458495948,
      "grad_norm": 1158.7845458984375,
      "learning_rate": 4.891867336123273e-05,
      "loss": 117.6687,
      "step": 22840
    },
    {
      "epoch": 0.18463303678924362,
      "grad_norm": 850.7025756835938,
      "learning_rate": 4.891662087938002e-05,
      "loss": 125.4999,
      "step": 22850
    },
    {
      "epoch": 0.18471383899352775,
      "grad_norm": 1056.5823974609375,
      "learning_rate": 4.891456649459155e-05,
      "loss": 104.6559,
      "step": 22860
    },
    {
      "epoch": 0.1847946411978119,
      "grad_norm": 821.6203002929688,
      "learning_rate": 4.891251020703078e-05,
      "loss": 104.2007,
      "step": 22870
    },
    {
      "epoch": 0.184875443402096,
      "grad_norm": 842.5684814453125,
      "learning_rate": 4.8910452016861316e-05,
      "loss": 149.866,
      "step": 22880
    },
    {
      "epoch": 0.18495624560638013,
      "grad_norm": 1477.5130615234375,
      "learning_rate": 4.890839192424692e-05,
      "loss": 118.6691,
      "step": 22890
    },
    {
      "epoch": 0.18503704781066427,
      "grad_norm": 679.1465454101562,
      "learning_rate": 4.89063299293515e-05,
      "loss": 113.9977,
      "step": 22900
    },
    {
      "epoch": 0.1851178500149484,
      "grad_norm": 1265.08349609375,
      "learning_rate": 4.890426603233913e-05,
      "loss": 187.4383,
      "step": 22910
    },
    {
      "epoch": 0.18519865221923254,
      "grad_norm": 1642.4959716796875,
      "learning_rate": 4.890220023337402e-05,
      "loss": 114.8031,
      "step": 22920
    },
    {
      "epoch": 0.18527945442351668,
      "grad_norm": 1289.17529296875,
      "learning_rate": 4.890013253262052e-05,
      "loss": 139.2596,
      "step": 22930
    },
    {
      "epoch": 0.1853602566278008,
      "grad_norm": 1148.701171875,
      "learning_rate": 4.889806293024317e-05,
      "loss": 115.5268,
      "step": 22940
    },
    {
      "epoch": 0.18544105883208495,
      "grad_norm": 854.2969970703125,
      "learning_rate": 4.889599142640663e-05,
      "loss": 104.5058,
      "step": 22950
    },
    {
      "epoch": 0.18552186103636908,
      "grad_norm": 638.8955078125,
      "learning_rate": 4.889391802127572e-05,
      "loss": 137.5948,
      "step": 22960
    },
    {
      "epoch": 0.1856026632406532,
      "grad_norm": 1359.7989501953125,
      "learning_rate": 4.8891842715015415e-05,
      "loss": 155.3802,
      "step": 22970
    },
    {
      "epoch": 0.18568346544493733,
      "grad_norm": 900.0406494140625,
      "learning_rate": 4.888976550779082e-05,
      "loss": 148.2721,
      "step": 22980
    },
    {
      "epoch": 0.18576426764922147,
      "grad_norm": 950.3909912109375,
      "learning_rate": 4.888768639976723e-05,
      "loss": 128.0501,
      "step": 22990
    },
    {
      "epoch": 0.1858450698535056,
      "grad_norm": 904.05615234375,
      "learning_rate": 4.888560539111007e-05,
      "loss": 117.1624,
      "step": 23000
    }
  ],
  "logging_steps": 10,
  "max_steps": 123750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}